Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
Int J Adv Manuf Techno1 DO1 10.1007/s00170-0124399-2
Pattern recognition for bivariate process mean shifts using feature-based artificial neural network
Ibrahim Masood . Adnan Hassan
Rwei~rd. I July 2U10 Acuepr~d: I6 July 2012 1: Syrmp;r-\'crlag Lunllon l.irniicd 2012
Abstract In multivariate quality control, the rutificial neu- ral networks (ANN)-based pattem recognition schemes gen- erally performed better for monitoring bivariate process mean shifts and provided more efficient information for diagnosing the source variable(s) compared to the traditional multivariate statistical process control charring. However, these schemes revealed disadvantages in term of reference bivariate pattern in identifying the joint effect and exeess false alarms in identfyhg stable process condition. In this study, feature-based ANN scheme was investigated for rec- ognizing bivariate correlated patterns. Featurebasal input representation was utilized into an ANN training and te* towards strengthening discrimination capability between bi- variate normal and bivariate mean shift patterns. Besides indicating an effective diagnosis capability in dealing with low correlation bivariate pattern, the proposed scheme promotes a smaller network size and better monitoring ca- pability as compared to the raw data-based ANN scheme.
Keywords Artificial neural networks . Bivariate emelated pattem . Process monitoring and diagnosis . Statistical features . Pattern recognition
L Masood (B) Faculty of Mechanical and Maouficturing Engineering, Universiti Tun Hussein Onn Mdaysia, 86400 Pait Rajq Baru Pahat, Johor, Malaysia e-mail: ibrahim@uthmedumy URL. hnp://im.uthin.edu.my
A. Hassan Faculty of Mechanical Engineering, Universiti Teknologi Malaysia, 81310UTM Sbd& Johor, Malaysia e-mail: [email protected] WRC. http:l/w~\w.utm.rny
Abbreviations ANN Aaificial neural network ARL Average run length ART Adaptive resonance theory BPR Bivariate pattern recognition CCPR Control chart pattern recognition CUSUM Cumulative sum EWMA Exponentially weighted moving average FAR False alarm rate i.i.d. Identically and independently distributed LEEWMA Last value of exponentially weighted moving
average LVQ Learning vector quantization MAT (Mean) x (autocomelation) 1MGUSUM Multivariate cumulative sum MEWMA Multivariate exponentially weighted moving
average MLP Multilayer perceptrons M S V (Mean) x (mean square value) MQC Multivariate quality control MRWA Multi-resolution wavelet analysis MSD (Mean) x (standard deviation) MSE Mean square e m MSPC Multivariate statistical proeess control YLA k c i p l e component anaiysis RA Recognition a c c m y percentage RBF Wal basis function SOM Kohonen self-organizing mapping SPC Statistical process control traingdx Gradient decent with momentum and adaptive
leaming rate
1 Introduction
It is well known that variation in manufacluring processes has become a major source of poor quality. W~ear and tear,
a springs Published online: 04 August 2012
Int I Adv Manuf Teehnal
For the future work, we plan to conduct a more extensive investigation towards heightening the performance of the
feature-based ANN for identifying smaller mean shifts. In-
vestigation also wili be extended for other causable pa t te rn
such as trend and cyclic.
Acknowledgments The authors would like to thank Universiti Tun Hussein OnnMalaysia, Universiti Tehologi Malaysia, and M b b t r y of Higher Education of a y s i a who are sponsoring this work The constructive suggations h m the Editor in Chief of UAMT and the anonymous reviewers have greatly aided us to jmprove the contmts as well as the presentation of this paper
References
1. Oakland JS (1996) Statistical process control. Butteworth- Heincmann, Oxford
2. Shewhad WA (1931) The eecanomjc wntml of quality m f a o tured products. Van Nostrand, New York
3. Nelson LS (1989) Standardization of Shewhart control charts. J Qual Techno1 21(4):287-289
4. Cmsier RB (1986) A new two-sided cumulative sum quality control schemes. Technometrics 28[3):187-194
5. Lucas JM, Saccucci MS (1990) Exponentially weighted moving averaee control schemes: urnoeaies and enhancements. Techno-
A .
rnetrii 32(1):1-12 6. Hotelling HH (1947) Multinkte aualiW control. In: Eisenhart C,
Hastay kW, W ~ U ~ ~ ' W A (eds) ~'hni~nes of statistical analysis McGraw-Hill New York
7. Cmsier R3 (19881 Multivariate eeneralizations of cumulative sum - quality coniol schemes. Technornehics 30(3]:291-303
8. P~gnafieUo U, R u m GC (1990) Conpansom of multivariate CIJSUM ch&s. J Qual Techno1 22(3):173-186
9. Lowry CA, Wwdall WH, Champ CW, Rigdon SE 11992) M a - variate exponentially weightedrnoving average control chart. Technomdcs 34(1):46-53
10. Prabhu SS, Runga GC (1997) Designing a multivaziate EWU4 control chart I Qual Teehnal 29(1):8-I5
11. Alt FB (1985) Multivariate quality control. In: Johnson NL, Kotz S (eds) Encyclopedia of statistical sciences 6. Wiley, New York
12. Jwkson JE (1991) A user's guide to principle component. Wiley. New Jersey
13. Fuch C, Benjamini Y (1994) Multivariate profile c h w for sta- tistical oroms controL Technamehim 36i231182-195
14. ~ a s a n - k , Tracy ND, Young JC (1995) D&mpositian of 15 for mulhmate eontrol chart mntemrefauon. J W l - m P .. . 108
15. Mason RL, Tracy ND, Young JC (1996) Monitoring a multivariate SteD DrOCeSS. J Oual Te~6nol2811):39-50 . .
16 Srpnil\cda A. N.xhl..r JA (199-J A ..in~ulauon apprdnch lo multi- \dndtc~sualit) c.>nrrol. ('urnpat lnd Fne 33(1- 2):l 13-1 lo
17. Lowry CA Montgomery DC ( 1 9 9 5 ) ~ review of multivariate eontror charts. llE Tmos 27[6):81)(31810
18. KO& T. MacGreeor SE (1996) Multivariate SPC methods for . . proce,.; 3nJ prdu ;? ,non!lonng. J C)L.IJ 1'cchnl,l 2P(1):J1I1J426
IQ. Mxwn 1U. Tnry ND, Yourl* Ji t1YL)7) A ~racti;~l appruach lur inte'preiing mliivariate f control chart &n&. J Q&I Techno1 29(4):39W06
20. Bersimis S, Psarakis S. Panaretos J (2007) Multivariate statistical process control charts an o-iew. ~ual'&liab EngInt23:517- 543
21. W6hler C, Aolauf JK (1999) A time delay neural network algo- rithm for &mating image-pattern shape and motion. Image Vis - - Comput 17-281-294
22. Singh V, Mishra R (2006) Development a machine vision system for spangle classification using image processing and artificial neural nehvmk Surf Coat Technol201:2813-2817
23. Liang SF, Lu SM, Chang JY, Lin CT (2008) A novel two-stage imoulse noise removal echniaue based on n e d networks aid fu& decision. IEEE T m F& Syst 16(4):863-873
24. Chdstodoulou CI, Pattichis CS (1999) UnsuDenised ~at tem rec- ognition for the classificationof &G Signals. ~ E E Trans Bimned Eng 46(2):16%178
25. Giiven A, Kara S (20061 Diagnosis of the macular diseases fiom pattern electroretinogmphy signals using artificial neural netwo~ks. Expal Svst ADD] 30:361-366
26. ~ l t u n s ~ S, Teiatarz, Emrogul 0, Aydur E (2009) Anew approach to uninary system dynamics problems: evaluafion and classification of uroflo&neter iienals usine artificial neural networks. Exoert - -
Syst Appl36:4891>895 27. Hr 0, Yumusak N, Temvrtas F (2010) Chest diseases d~agnosis . .
using artificial neural networks. Expert Syst Appl 37:764b 7 K C C , w--
28. Chen CW (2009) Modeling and control for nonlinear structural systems via a M*T-based approach. Expert Syst Appl 36:4765- 4772
29. Cheo CW, Yeh K, Liu KFR (2009) Adaptive iiizzy sliding mode contml for seismically exited bridges mth lead rubber beariDg isolation. Int J Unce~tain Fuzziness Knowl-Based Syst 17(5):705- 777
30. Chen CY. Lin JW, Lee WL Chen CW 12010) F ~ w control for an wcdnic ruucrwi.. 3 cfis: S R I . ~ ~ in rirrl:-dsIay 1'1 I' ~ I I C ~ . J Vib C.unllol lh(lJ.147 Ifll
31. Amin A (2000) Rewenition of minted Arabic text based on elobal . , - feahlres and decision tree learning techniques. Pattern Recogu 331309-1323
32. BhaUacMa K, Sarma KK (2009) ANN-based innovative seg- mentation method for handwritten text in assam-. Int J Cornput Sci Issues 5 9 1 6
33. Marcu T, Seliger BK, Stiicher R (2008) Design of fault detection for ahydmnlic looper using dynamic neural networks. Conhol Eng Pract 16:192-213
34. Demetgul M, T m e l IM, Taskin S (2009) Fault diagnosis of oneumatic svstems with ardficial neural network alearithms. Ex- k n Sknt h i 1 36:11)512 -1U51Y
35. Ji.4 LY, Ma JN', Nang FJ, l.iu N (ZUIG, ('h~r.~;tr.riu~;, t'.,r.x?slicc of hydraulic valve based on grey ro~~elation and AtWS EX~G Syst Appl37:125&1255
36. H a v k S (1999) Neural networks: a commehemive foundation PAtice H& N ~ W Jersey
37. Hwarng HB, Hubele NF (1993) Back-propagation pattern reeog- - E k i + h f 2 G b ~ & ~ c m e M ~ -
Comput Ind Eng 24:219-235 38. V e l w T (1993) Back propagation aaii~cial neuralnetworks for
the analysis of quality control c h a t s Comput Ind Eng 25(L- 4):397-400
39. Pham DT, Oztemel E (1993) Control chart pattem rewgnition using combinations of muhilayer pemptroos and leaning vector qwanthtion -neural nelworks. Pros Inst Mecb Eng 207: 11 3-1 18
40. Pham DT, Oztemel E (1994) Control chart pattem r-gnition using learning vector quantization networks. In1 J Prod Res 32721-729
41. Zomassafine F. Tanneck JDT 11998) A review of n& oetwmks . , for statistical pmcess control. J Intell Manuf 9:209-224
42. Masood I, Hawan A (2010) Issues in developmes of artificial neural network-based control chart patfern r&om schemes. Eur 3 Sci Res 39(3):33M55
shier Goin n.ubivrise P e h u l \ig.nuls. Co~nput Inti Fng 47:195--205 24. Niaki SIX, Ahbh 3 (2~05) I:3ult disgfiusli in rnn~l~lvnrjarr. c711-
-01 charts ushe mircial neural nehvorks. Oual Reliab Ene Int . - 2 1 :82&840
45. Cbeng CS, Cheng HP (2008) Identifying the saurce ofvariance shifts in the multivariate process using neural networks and sup- port vector machines. Expert Syst Appl35:19&206
46. Yu JB. Xi LF. Zhou XI (2009) ldentifvine source(^) of out-of- . , > - , , control signals in multivariate m a n u b r i n g prmesses using se- lective new1 nchvork ensemble. Ene Avvl Artif Intell 22141-152
iate pracem. ~ o r n ~ u t Ind Eng 4 4 3 8 5 4 8 48. G u h RS (2007) On-line identification and quantification of mean
shifm in bivariate processes using a NN-bmed approach. Qua1 Reliab Eng Int 23:367-385
49. Yu W, Xi LF (2W9) A neural network ensemble-based model for on-line monitoring and diagnosis of out-ofantto1 signals in mul- tivariate manufacturing processes. Expert Syst Appl36:909-921
50. EI-Midany TT, El-Baz MA, Abd-Elwahed US (2010) A nrmsed . . f i i w w u r l ~ L I cdnnc l rlj.ln pmcni recagn~lhrr in mullivlrifi plo- cci, winp ~niricljl ~new.il~~r.tworks. I.xpm S)a Appl37:1035 ,012
51 Park S-J. L e \I-$. Shir. S-Y. Chr, K-H. Lint J- 1. Cho U-S. Y-H ~ ~
Jci M-K, park G H (2005) ~un-&-mn overlay eo&l of stqpers in semiconductor m m i a d g systems based on hitor- ical data analysis and neural network modeling. IEEE Tram Semi- con Manuf 18(4):60%13
52. hlnnm RL, Chou 1%. Yani: JC {2lIJI) .Applying 1Llr.lling 7' slatirrs w hitch prorrssc,. J Qua1 T<ihnr,l 33(4):40M7:,
j3. Parrd h.l(iL, L I M J ~ !'K (20~3-7U01, Ap~lkdtion of nul1iv~vi;;le , ., control chai and ~a&n-Tmcy dmmpasition procedure to the
study of the consistency of impurity profiles of drug substances. Oualitv Eneineerine 16(1):127-142
54 &gbk &lbIeathth~ b b 2 ) A real-time information system far mulfivariatte staiisticalprffiess contml. lot JProdEeon 75161-172
55. Miletic I, Quinn S, ~ i d d c M, Vaculik V, Champagne M (2004) An industrial perspedive on implementation on-line applications o f multivariate statistics J Pmeess Control 14:821-836
5 6 Lehman RS (19771 Computer simulation and m o d e l i i an Intro- duction. Lawrence Erlbaum, London
57. B m e U H, Eriksson J (2000) Feature reduction for classification o f multidimensional data Pattern Recog 33(12):1741-1748
58. Klosgen W, Zytkow JM (2002) Handbook of data mining and knowledge discovery. Oxford University Press, London
In1 i Adv Manuf Techno1
59. Pham DT, Wani MA (1997) F a r e - h e d wm01 c M partem rewpition IatJ Prod Res 35(7):1875-1890
60. Gauri SK Chakrabor@ S (2006) Feature-based recognition of control chart patterns. Comput Ind Eng 51:72&742
61. Gaui SK, Chakrabmty S (21248) b r o v e d recopnition oicontrol chchart paltern using artificial neuafnetworks. &t J Rdv kf Techno1 36:1191-1201
62. Hassan A, Nabi Baksh MS, Shaharoun MA, Jamaludin H (2003) Improved SPC chart pattern recognition using sfatistical featllres. Int JPrcd Res 41(7):1587-1643
63. Guh RS (2010) Simultaneous pmcess mean and variance monitor- ing using artificial neural networks. Comput Ind Eng 58:739-753
64. A1-hsaf Y (2004) Recognition of control chart pan- using multi-resolution wavelefs analysis and neural corks. Comput Ind Eng 47:17-29
65. Assaleh K, Al-AssafY (2005) Features extraction and analysis for classifying musable pattern in control charts. Comput Ind Eng 49:168-181
66. Chen Z, Lu S, Lam S (2007) A hybn'd system for SPC con-nt pattem recognition. Adv Eng Inform 21:30>310
67. Wang CH, Kuo W, Qi H (2W7) An integrated approach forproms monitoring ming wavelet analysis and competitive neural netwn~k In t i Prod Res 45:227-244
68. Montgomery DC (2005) Introduction to statistical quality control. Wiley, New Jersey
69. Demuth H, Beale M (1998) Neural network twlbox, user's guide. The Mathworks, Natick
70. Bishop C (1995) Neural nehvork for pattern recognition. Oxford University Press, New York
71. Cheng CS (1995) A multi-layer neural network model for detect- ing changes in the process meaa Comput Ind Eng 2 8 5 6 1
72. Cheng CS (1997) A mural network appraacb i m the analysis of contml chart pattern. Int Pmd Res 35(3):667497
73. Guh RS, Hsieh YC (1999) A neural network based model for abnormal pattern recognition af control charts. Comput Ind Eng 3697-108
74. Guh RS, Tannock JDT, O'Brien C (1999) Intellispc: a hybrid intelligence tool for on-line economical statistical process mntml. Expert Syst Appl17: 19S2 12
75. Guh RS, Z o h t i n e 8, T w m k JDT, O'Brien C (1999) On-line control cbm pattern detection and discriminatio-a neural net- work approach. Artif Intel1 Eng 13:41%425
76. P e w MB, Spoerre JY Velasco T (2001) Conmi chart pattern recognition using back propagation df ic ia l neural networks. Int J Prod Res 39:3399-3418