Kussul / Baidyk Neural Networks and Micromechanics

1. Auflage 2009
ISBN: 978-3-642-02535-8
Verlag: Springer
Format: PDF
Kopierschutz: Wasserzeichen (»Systemvoraussetzungen)

E-Book, Englisch, 221 Seiten

Micromechanical manufacturing based on microequipment creates new possibi- ties in goods production. If microequipment sizes are comparable to the sizes of the microdevices to be produced, it is possible to decrease the cost of production drastically. The main components of the production cost - material, energy, space consumption, equipment, and maintenance - decrease with the scaling down of equipment sizes. To obtain really inexpensive production, labor costs must be reduced to almost zero. For this purpose, fully automated microfactories will be developed. To create fully automated microfactories, we propose using arti?cial neural networks having different structures. The simplest perceptron-like neural network can be used at the lowest levels of microfactory control systems. Adaptive Critic Design, based on neural network models of the microfactory objects, can be used for manufacturing process optimization, while associative-projective neural n- works and networks like ART could be used for the highest levels of control systems. We have examined the performance of different neural networks in traditional image recognition tasks and in problems that appear in micromechanical manufacturing. We and our colleagues also have developed an approach to mic- equipment creation in the form of sequential generations. Each subsequent gene- tion must be of a smaller size than the previous ones and must be made by previous generations. Prototypes of ?rst-generation microequipment have been developed and assessed.

Autoren/Hrsg.

Kussul, Ernst

Baidyk, Tatiana

Inhaltsverzeichnis

1;FM;1
2;1;10
2.1;Chapter 1: Introduction;10
2.1.1;References;13
3;2;15
3.1;Chapter 2: Classical Neural Networks;15
3.1.1;Neural Network History;15
3.1.2;McCulloch and Pitts Neural Networks;15
3.1.3;Hebb Theory;16
3.1.4;Perceptrons;18
3.1.5;Neural Networks of the 1980s;20
3.1.6;Modern Applications of Neural Network Paradigms;23
3.1.6.1;Hopfield Neural Networks;23
3.1.6.2;Adaptive Resonance Theory (ART);24
3.1.6.3;Self-Organizing Feature Map (SOFM) Neural Networks;25
3.1.6.4;Cognitron and Neocognitron;26
3.1.6.5;Backpropagation;27
3.1.6.6;Adaptive Critic Design;28
3.1.7;RTC, RSC, LIRA, and PCNC Neural Classifiers;29
3.1.8;References;29
4;3;34
4.1;3: Neural Classifiers;34
4.1.1;RTC and RSC Neural Classifiers for Texture Recognition;34
4.1.1.1;Random Threshold Neural Classifier;36
4.1.1.2;Random Subspace Classifier;38
4.1.1.3;Encoder of Features;39
4.1.2;LIRA Neural Classifier for Handwritten Digit Recognition;40
4.1.2.1;Rosenblatt Perceptrons;41
4.1.2.2;Description of the Rosenblatt Perceptron Modifications;42
4.1.2.2.1;Mask Design;44
4.1.2.2.2;Image coding;45
4.1.2.2.3;Training procedure;45
4.1.2.2.4;Recognition procedure;47
4.1.3;LIRA-Grayscale Neural Classifier;48
4.1.4;Handwritten Digit Recognition Results for Lira-binary;49
4.1.5;Handwritten Digit Recognition Results for LIRA-Grayscale;50
4.1.6;Discussion;51
4.1.7;References;52
5;4;54
5.1;Chapter 4: Permutation Coding Technique for Image Recognition System;54
5.1.1;Special- and General-Purpose Image Recognition Systems;54
5.1.2;Random Local Descriptors;56
5.1.3;General Purpose Image Recognition System Description;57
5.1.4;Computer Simulation;61
5.1.5;Permutation Coding Neural Classifier (PCNC);61
5.1.5.1;PCNC structure;61
5.1.5.2;Feature extractor;62
5.1.5.3;Encoder;63
5.1.6;PCNC Neural Classifier Training;71
5.1.7;Results Obtained on the MNIST Database;72
5.1.8;Results Obtained on the ORL Database;73
5.1.9;References;78
6;5;81
6.1;Chapter 5: Associative-Projective Neural Networks (APNNs);81
6.1.1;General Description of the Architecture;81
6.1.1.1;Neuron, the Training Algorithms;81
6.1.1.2;Neural Fields;84
6.1.2;Input Coding and the Formation of the Input Ensembles;89
6.1.2.1;Local Connected Coding;89
6.1.2.1.1;Coding the numbers and sets of the numerical parameters;91
6.1.2.1.2;Code normalization;92
6.1.2.2;Shift Coding;94
6.1.2.2.1;Centering the shift code;98
6.1.2.2.2;Application of shift coding;102
6.1.2.3;Functions of Neural Ensembles;102
6.1.2.4;Methods of Economical Presentation of the Matrix of Synaptic Weights (Modular Structure);104
6.1.2.4.1;Stochastic not fully connected networks;104
6.1.2.4.2;Constructing modular neural networks;107
6.1.3;Conclusion;108
6.1.4;References;109
7;6;111
7.1;Chapter 6: Recognition of Textures, Object Shapes, and Handwritten Words;111
7.1.1;Recognition of Textures;111
7.1.1.1;Extraction of Texture Features;111
7.1.1.2;The Coding of Texture Features;112
7.1.1.3;Texture Recognition;113
7.1.1.4;The Experimental Investigation of the Texture Recognition System;115
7.1.1.5;Texture Recognition with the Method of Potential Functions;117
7.1.2;Recognition of Object Shapes;119
7.1.2.1;Features for Complex Shape Recognition;120
7.1.2.2;Experiments with Algorithms of Complex Shape Recognition;122
7.1.3;Recognition of Handwritten Symbols and Words;123
7.1.3.1;Features Utilized for the Recognition of Handwritten Words;123
7.1.3.1.1;The algorithm of line thinning;124
7.1.3.1.2;Algorithm of line thickening;125
7.1.3.1.2.1;Extraction of informative features;126
7.1.3.1.2.2;Information coding;126
7.1.3.1.2.3;Coding of binary features;127
7.1.3.1.2.4;Coding the feature position on the image;127
7.1.3.1.2.5;The experimental results;129
7.1.3.1.2.6;Optimization of the feature set;129
7.1.4;Conclusion;133
7.1.5;References;134
8;7;136
8.1;Chapter 7: Hardware for Neural Networks;136
8.1.1;Neurocomputer NIC;136
8.1.1.1;Description of the Block Diagram of the Neurocomputer;136
8.1.1.2;The Realization of the Algorithm of the Neural Network on the Neurocomputer;138
8.1.2;Neurocomputer B-512;139
8.1.2.1;The Designation of the Neurocomputer Emulator;140
8.1.2.2;The Structure of the Emulator;140
8.1.2.3;The Block Diagram of B-512;143
8.1.3;Conclusion;145
8.1.4;References;145
9;8;146
9.1;Chapter 8: Micromechanics;146
9.1.1;The Main Problems of Microfactory Creation;146
9.1.2;General Rules for Scaling Down Micromechanical Device Parameters;150
9.1.3;The Analysis of Micromachine Tool Errors;153
9.1.3.1;Thermal Expansion;153
9.1.3.2;Rigidity;155
9.1.3.2.1;Compression (or extension) of the bar;155
9.1.3.2.2;Bending of the bar. Case 1;156
9.1.3.2.3;Bending of the bar. Case 2;157
9.1.3.2.4;Torsion of the bar;157
9.1.3.3;Forces of Inertia;158
9.1.3.3.1;Force of inertia. Linear movement with uniform acceleration;158
9.1.3.3.2;Centrifugal force;159
9.1.3.4;Magnetic Forces;160
9.1.3.5;Electrostatic Forces;160
9.1.3.6;Viscosity and Velocity of Flow;161
9.1.3.6.1;Pneumatic and hydraulic forces;161
9.1.3.6.2;Forces of surface tension (capillary forces);163
9.1.3.7;Mass Forces;164
9.1.3.8;Forces of Cutting;164
9.1.3.9;Elastic Deformations;166
9.1.3.10;Vibrations;167
9.1.4;The First Prototype of the Micromachine Tool;169
9.1.5;The Second Prototype;172
9.1.6;The Second Micromachine Tool Prototype Characterization;174
9.1.6.1;Positional Characteristics;175
9.1.6.2;Geometric Inspection;182
9.1.7;Errors that Do Not Decrease Automatically;183
9.1.7.1;Methods of Error Correction;183
9.1.7.1.1;The method of ``lever.´´;184
9.1.7.1.2;Micromachining center based on parallograms;185
9.1.7.1.3;Parallel micromanipulators;185
9.1.8;Adaptive Algorithms;186
9.1.8.1;Adaptive Algorithms Based on a Contact Sensor;186
9.1.9;Possible Applications of Micromachine Tools;190
9.1.9.1;The Problem of Liquid and Gas Fine Filtration;190
9.1.9.2;Design of Filters with a High Relation of Throughput to Pressure Drop;191
9.1.9.3;An Example of Filter Design;193
9.1.9.4;The Problems of Fine Filter Manufacturing;193
9.1.9.5;The Filter Prototype Manufactured by the Second Micromachine Tool Prototype;194
9.1.9.6;Case Study;194
9.1.10;Conclusion;196
9.1.11;References;197
10;9;200
10.1;Chapter 9: Applications of Neural Networks in Micromechanics;200
10.1.1;Neural-Network-Based Vision System for Microworkpiece Manufacturing;200
10.1.2;The Problems of Adaptive Cutting Processes;201
10.1.3;Permutation Coding Neural Classifier;202
10.1.3.1;Feature Extractor;202
10.1.3.2;Encoder;204
10.1.3.3;Neural Classifier;206
10.1.4;Results;207
10.1.5;References;208
11;10;209
11.1;Chapter 10: Texture Recognition in Micromechanics;209
11.1.1;Metal Surface Texture Recognition;209
11.1.2;Feature Extraction;211
11.1.3;Encoder of Features;211
11.1.4;Results of Texture Recognition;212
11.1.5;References;213
12;11;214
12.1;Chapter 11: Adaptive Algorithms Based on Technical Vision;214
12.1.1;Microassembly Task;214
12.1.2;LIRA Neural Classifier for Pin-Hole Position Detection;219
12.1.3;Neural Interpolator for Pin-Hole Position Detection;220
12.1.4;Discussion;223
12.1.5;References;224