Development of prediction systems using artificial neural networks for intelligent spinning machines [Elektronische Ressource] / Farooq, Assad

Description

Development of Prediction Systems Using Arti cial NeuralNetworks for Intelligent Spinning MachinesVon der Fakult at MaschinenwesenderTechnischen Universit at DresdenzurErlangung des akademischen GradesDoktoringenieur (Dr.-Ing.)angenommene DissertationM. Sc. Farooq, Assadgeb. am 19.11.1977 in Faisalabad PakistanTag der Einreichung: 26.01.2010Tag der Verteidigung: 06.05.2010Gutachter: Prof. Dr.-Ing. habil. Dipl.-Wirt. Ing. Chokri CherifProf. Dr.-Ing. Burkhard WulfhorstProf. Dr.-Ing. habil. Thorsten SchmidtVorsitzender der PromotionskommissionAcknowledgmentsI am indebted to many people for the successful completion of this dissertation. First ofall, I owe my deepest gratitude to my supervisor, Prof. Dr.-Ing. habil. Dipl.-Wirt. Ing.Chokri Cherif, for the support and con dence that he has given to me. His vision, ideasand comments on various issues have contributed to the quality of this dissertation.It is an honor for me to thank Prof. Dr. -Ing. Burkhard Wulfhorst, for his acceptance toreferee this dissertation and for giving the valuable suggestions to improve my work.I would like to express my sincere gratitude and appreciation for Dr. Ing. Thomas Pusch,He has been with me through out my Ph.D. as mentor, colleague and editor of this disser-tation. His invaluable advice, continuous suggestions, encouragement and constructivecriticism has made this work possible.

Informations

• Publié par: technische_universitat_dresden
• Publié le: 01 janvier 2010
• Nombre de lectures: 18
• Langue: English
• Poids de l'ouvrage: 13 Mo

Extrait

Development of Prediction Systems Using Artificial Neural Networks for Intelligent Spinning Machines

Tag der Einreichung: Tag der Verteidigung:

Gutachter:

Von der Fakultät Maschinenwesen

der

Technischen Universität Dresden

zur

Erlangung des akademischen Grades

Doktoringenieur (Dr.Ing.)

angenommene Dissertation

26.01.2010 06.05.2010

M. Sc. Farooq, Assad

geb. am 19.11.1977 in Faisalabad Pakistan

Prof. Dr.Ing. habil. Dipl.Wirt. Ing. Chokri Cherif

Prof. Dr.Ing. Burkhard Wulfhorst

Prof. Dr.Ing. habil. Thorsten Schmidt

Vorsitzender der Promotionskommission

Acknowledgments

I am indebted to many people for the successful completion of this dissertation. First of all, I owe my deepest gratitude to my supervisor, Prof. Dr.Ing. habil. Dipl.Wirt. Ing. Chokri Cherif, for the support and confidence that he has given to me. His vision, ideas and comments on various issues have contributed to the quality of this dissertation.

It is an honor for me to thank Prof. Dr. Ing. Burkhard Wulfhorst, for his acceptance to referee this dissertation and for giving the valuable suggestions to improve my work.

I would like to express my sincere gratitude and appreciation for Dr. Ing. Thomas Pusch, He has been with me through out my Ph.D. as mentor, colleague and editor of this disser tation. His invaluable advice, continuous suggestions, encouragement and constructive criticism has made this work possible. Also I would like to thank my research group fellows for their support, it has been a pleasure working with them. I also like to pay my gratitude to all the employees of Institute of Textile Machinery and High Performance Material Technology, who have made available their support in a number of ways. Espe cially, Mrs. Stohr, Mr. Berndt and Mr. Posselt for helping to carry out the experimental part of this research work. I also like to say thanks to students who worked with me for timely completion of this dissertation.

I am also grateful to Higher Education Commission of Pakistan (HEC) for providing the financial support for this dissertation and German Academic Education Exchange (DAAD) for their extended administrative support.

I am thankful to Rieter Ingolstadt GmbH, for providing me an opportunity to facilitate my research work by providing experimental machines and materials. Especially, Mrs. Corina Wiede for the technical support during my stay in Ingolstadt.

I would also like to extend the deep emotions of appreciation and love to my family members, aunt Salma, brother Engr. Ahmad Farooq his wife and children, Beenish and Huzefa, and to my friends and all those hands lifted to pray for me. A very special acknowledgement to my mother, for being first teacher of my educational career. The foundation laid by her helped me at every stage until the completion of this dissertation. Also for her love, sympathetic support, special prayers and best wishes.

Finally, I dedicate this dissertation to the loving memories of my late father Prof. Dr. Sardar Ali.

Dresden, 25.01.2010

Assad Farooq

10

Tasks of Draw Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

8

1

5

AutoLeveling . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

10

5

viii

SymbolsandAbbreviations

SignificanceofDrawFrame

Introduction

Doubling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Importance of Draw Frame in Spinning Process

Influence of Draw Frame on Following Textile Processes . . . . . . . . . .

. . . . . . . . . . . . . .

14

11

2.3.1

Contents

State of the Art

2

2.3.3

Drafting Process

2.3.6

3.1.1.1

Ideal Drafting versus Real Drafting . . . . . . . . . . . .

2.3.5

2.3.2

2.3.4

3

3.1

3.1.1

Blending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12

Parallelization . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

13

Optimization of Draw Frame . . . . . . . . . . . . . . . . . . . . . . . . .

21

19

2.2

2.3

2.1

19

. . . . . . . . . . . . . . . . . . . . . . . . . . .

19

Cleaning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Drafting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

17

57

. . . . . . . . . . . . . . .

6.1

6.2

. . . . . . . . . . . . . . . .

37

Material Selection . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

48

49

5

5.1

35

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Machine Optimization . . . . . . . . . . . . . . . . . . . . . . . .

25

Material Selection . . . . . . . . . . . . . . . . . . . . . . . . . . .

Table of Contents

5.4

Drafting Theories . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3.2

Ring Spun Yarns

3.1.2

ii

. . . . . . . . . . . . . . . . . . . . . . . .

46

41

45

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Prediction Modeling

3.3

27

38

Comparison of Neural Network with Other Models . . . . . . . .

36

Evenness, Imperfections and Hairiness

Sliver Cohesion . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

32

52

56

61

61

63

57

58

Investigations on Leveling Action Point . . . . . . . . . . . . . . . . . . .

48

52

A Brief History of the Field

6

Artificial Neural Networks

5.3

5.3.1

5.4.1

Yarn Manufacturing

4

Objective of the Research

Materials and Methods

5.2.1

5.2.2

5.2

3.3.3

5.4.2

5.4.3

3.3.2

3.3.1

Testing Plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Rieter Draw Frame RSBD40

Application to Yarn Manufacturing

Yarn Properties Prediction . . . . . . . . . . . . . . . . . . . . . .

Autoleveling

3.1.1.2

Investigations on Sliver Quality Characteristics . . . . . . . . . . . . . . .

83

. . . . . . . . . . . . . . . . . .

82

. . . . . . . . . . . . . . . . . . . . . . .

81

Classification of ANN . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6.8.4.2

6.8.4.1

. . . . . . . . . . . . . . . . . . . . . . . . .

Table of Contents

6.3

6.4

. . . . . . . . . . . . . . . . . . . . . . . .

77

Underfitting and Overfitting . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

80

Optimization of Network Parameters . . . . . . . . . . . . . . . .

80

Network Initialization

77

. . . . . . . . . . . . . .

79

78

74

Feedforward Pass . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . .

74

Momentum . . . . . . . . . . . . . . . . . . . . . . . . .

ANN Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

6.8

Training a Neural Network . . . . . . . . . . . . . . . . . . . . . . . . . .

Error Surface . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . .

iii

Exclusive OR /(XOR) Problem . . . . . . . . . . . . . . . . . . . . . . .

68

6.8.5.3

6.8.5.2

6.8.5.1

71

72

73

LevenbergMarquardt Technique

Learning Rate

71

Recurrent Networks & Networks having Shortcut Connections . .

70

6.8.6

70

64

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

73

6.8.5

Network Structure . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Biological Inspiration

. . . . . . . . . . . . . . . . . . . . . . .

6.8.1

Input Selection

Backward pass

Calculation of Error

Back Propagation

80

79

Problems During Training . . . . . . . .

6.7.3

6.7.1

6.7.2

6.7

Feedforward Network

6.6.1

6.5.2

6.5.1

6.6

6.8.2

6.8.3

6.5

6.8.4

Supervised Learning

Unsupervised Learning

6.6.2

Preprocessing of Data . . . . . . . . . . . . . . . . . . . . . . . .

77

75

90

95

6.12 Software Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Table of Contents

Early Stopping . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

7.1.4

7.1.3

Comparison Between Polyester and Viscose

6.8.7.2

Regularization

Testing A Trained Network

. . . . . . . . . . . . . . . . . . . . . . .

7

6.8.7.1

Generalization

84

84

iv

6.8.7

6.9

89

84

7.1.7

Comparison Among Materials . . . . . . . . . . . . . . . . . . . . . . . . 110

Infeed Variations

Polyester/Cotton Blends . . . . . . . .

Training and Test Performance of Neural Networks

7.1.1

LAP Influencing Parameters . . . . . . . . . . . . . . . . . . . . . . . . .

Leveling Action Point

6.11 Conclusion . . . . . . .

7.1.6

7.1.5

7.2

7.2.1

7.2.2

7.4.1

7.4.2

7.3

7.1.2

Feeding Speed . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

96

. . . . . . . . . . . . . 119

Selection of Relevant Input Parameters . . . . . . . . . . . . . . . 117

6.10 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . 103

. . . . . . . . . . . . . . . 113

. . . . . . . . . . . . 111

Break Draft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102

Sliver Deflection Bars . . . . . . . . . . . . . . . . . . . . . . . . . 105

7.4

7.1

. . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . 117

88

86

. . . . . . . . . . . 108

Break Draft Distance . . . . . . . . . .

. . . . . . . . . . . . . . . .

95

Infeed Tension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100

. . . . . . . . . . . . . . . . . . . . . . . . . . . .

Analysis Using Artificial Neural Networks

Main Draft Distance . . . . . . . . . . . . . . . . . . . . . . . . . 104

Multiple Linear Regression Analysis . . . . . . . . . . . . . . . . . . . . . 114

Cotton Carded . . . . . . . . . . . . . . . . . . . . . . . 127

Polyester/Cotton 50/50 2nd Passage . . . . . . . . . . . 125

Training and Test Performances of Neural Networks . . . . . . . . . . . . 143

8.2

Selection of Relevant Input Parameters . . . . . . . . . . . . . . . 143

Effect of Doublings . . . . . . . . . . . . . . . . . . . . . . . . . . 140

Cotton Combed . . . . . . . . . . . . . . . . . . . . . . . 127

v

. . . . . . . . . 120

Effect of Main Draft Distance . . . . . . . . . . . . . . . . . . . . 139

Effect of Break Draft . . . . . . . . . . . . . . . . . . . . . . . . . 137

8

Effect of Delivery Speed . . . . . . . . . . . . . . . . . . . . . . . 135

. . . . . . . . . . . . . . . . . . . 123

8.2.2.2

8.2.2.1

8.2.2

8.2.2.3

Analysis Using Artificial Neural Networks

8.1.2

8.1.1

8.2.1

Sliver Quality Influencing Parameters . . . . . . . . . . . . . . . . . . . . 134

Effect of Break Draft Distance . . . . . . . . . . . . . . . . . . . . 138

. . . . . . . . . . . . . 144

Interaction effects among influencing variables . . . . . . . . . . . 141

Cotton 2nd Passage . . . . . . . . . . . . . . . . . . . . 121

8.1.4

8.1.3

8.1.6

8.1.5

Performance of the Trained Networks Using Industrial Data . . . 127

Conclusion, Practical Applications and Future Pathways . . . . . . . . . 128

Sliver CV(3m)%

Table of Contents

Leveling Action Point Prediction Function ”NEUROset” . . . . . 130

133

Sliver CVm% . . . . . . . . . . . . . . . . . . . . . . . . 145

CV(1m)% . . . . . . . . . . . . . . . . . . . . . . . . . . 146

. . . . . . . . . . . . . . . . . . . . . . 147

Polyester 2nd Passage

Cotton 1st Passage . . . . . . . . . . . .

7.5.1

7.5

7.4.3.2

Analysis of Sliver Characteristics

7.4.2.3

7.4.2.4

7.4.3.1

7.4.2.1

7.4.2.2

8.1

7.4.3

vi

. . . . . . . . . . . . . . . . . . . . . . 167

Influence of Break Draft . . . . . . . . . . . . . . . . . . . . . . . 154

Table of Contents

Sliver Cohesion . . . . . . . . . . . . . . . . . . . . . . . 149

. . . . . . . . . . . . . . . . . . . . . . . 165

9

9.1

9.2

Yarn Quality Influencing Parameters . . . . . . . . . . . . . . . . . . . . 152

. . . . . . . . . . . . . 160

Influence of Break Draft Distance . . . . . . . . . . . . . . . . . . 155

. . . . . . . . . . . . . . . . . . . . . . . . 161

Miscellaneous Influences . . . . . . . . . . . . . . . . . . . . . . . 157

. . . . . . . . . . . . . . . . . . . . . . 163

8.2.2.4

Analysis of Yarn Quality Using Artificial Neural Networks . . . . . . . . 159

10.2 Concept for Intelligent Spinning Machines

10 Concept Development of Intelligent Spinning Machines

Yarn CV (1m)%

151

Analysis of Yarn Quality

10.1 Applications of Research . . . . . . . . . . . . . . . . . . . . . . . . . . . 171

171

. . . . . . . . . . . . . . . . . 173

9.1.5

9.2.1

9.3

Influence of Delivery Speed

9.2.2

9.1.4

9.1.3

9.1.2

Yarn CV (3m)%

9.2.2.3

9.2.2.2

9.2.2.1

9.2.2.6

9.2.2.5

9.2.2.4

Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

Yarn Tenacity . . . . . . . . . . . . . . . . . . . . . . . . 166

Analysis Using Artificial Neural Networks

Influence of Main Draft Distance

Yarn Elongation

Selection of Relevant Input Parameters . . . . . . . . . . . . . . . 159

. . . . . . . . . . . . . . . . . . . . . . 162

. . . . . . . . . . . . . . . . . . 156

Yarn CVm%

8.3

Conclusion

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

Yarn Hairiness

9.1.1

. . . . . . . . . . . . . . . . . . . . . 153

Table of Contents

11 Summary and Outlook

List of Figures

List of Tables

Bibliography

vii

179

183

189

191