Neural Network Programming with Java Create and

unleash the power of neural networks by

implementing professional Java code 1st Edition
Souza pdf download
https://textbookfull.com/product/neural-network-programming-with-
java-create-and-unleash-the-power-of-neural-networks-by-
implementing-professional-java-code-1st-edition-souza/

Download more ebook instantly today - get yours now at textbookfull.com

We believe these products will be a great fit for you. Click
the link to download now, or visit textbookfull.com
to discover even more!

Artificial Neural Networks with Java - Tools for

Building Neural Network Applications 1st Edition Igor
Livshin

https://textbookfull.com/product/artificial-neural-networks-with-
java-tools-for-building-neural-network-applications-1st-edition-
igor-livshin/

Learning Network Programming with Java 1st Edition

Reese Richard M

https://textbookfull.com/product/learning-network-programming-
with-java-1st-edition-reese-richard-m/

Java Network Programming Fourth Edition Harold Elliotte

https://textbookfull.com/product/java-network-programming-fourth-
edition-harold-elliotte/

Learning Java Beginning programming with java for

dummies First Edition John Bach

https://textbookfull.com/product/learning-java-beginning-
programming-with-java-for-dummies-first-edition-john-bach/
Learning Java An Introduction to Real World Programming
with Java Marc Loy

https://textbookfull.com/product/learning-java-an-introduction-
to-real-world-programming-with-java-marc-loy/

A Functional Approach to Java: Augmenting Object-

Oriented Java Code with Functional Principles 1st
Edition Ben Weidig

https://textbookfull.com/product/a-functional-approach-to-java-
augmenting-object-oriented-java-code-with-functional-
principles-1st-edition-ben-weidig/

Learning Java An Introduction to Real World Programming

with Java 5th Edition Marc Loy

https://textbookfull.com/product/learning-java-an-introduction-
to-real-world-programming-with-java-5th-edition-marc-loy/

Motivation Effort and the Neural Network Model Theodore

Wasserman

https://textbookfull.com/product/motivation-effort-and-the-
neural-network-model-theodore-wasserman/

Deep Learning with JavaScript: Neural networks in

TensorFlow.js 1st Edition Shanqing Cai

https://textbookfull.com/product/deep-learning-with-javascript-
neural-networks-in-tensorflow-js-1st-edition-shanqing-cai/
Neural Network Programming
with Java

Unleash the power of neural networks by implementing

professional Java code

Fábio M. Soares
Alan M.F. Souza

BIRMINGHAM - MUMBAI
Neural Network Programming with Java

Copyright © 2016 Packt Publishing

All rights reserved. No part of this book may be reproduced, stored in a retrieval
system, or transmitted in any form or by any means, without the prior written
permission of the publisher, except in the case of brief quotations embedded in
critical articles or reviews.

Every effort has been made in the preparation of this book to ensure the accuracy
of the information presented. However, the information contained in this book is
sold without warranty, either express or implied. Neither the authors, nor Packt
Publishing, and its dealers and distributors will be held liable for any damages
caused or alleged to be caused directly or indirectly by this book.

Packt Publishing has endeavored to provide trademark information about all of the
companies and products mentioned in this book by the appropriate use of capitals.
However, Packt Publishing cannot guarantee the accuracy of this information.

First published: January 2016

Production reference: 1060116

Published by Packt Publishing Ltd.

Livery Place
35 Livery Street
Birmingham B3 2PB, UK.

ISBN 978-1-78588-090-2

www.packtpub.com
 Credits

Authors Project Coordinator

Fábio M. Soares Kinjal Bari
Alan M.F. Souza
Proofreader
Reviewer Safis Editing
Saeed Afzal
Indexer
Commissioning Editor Hemangini Bari
Amarabha Banerjee
Graphics
Acquisition Editor Disha Haria
Rahul Nair
Production Coordinator
Content Development Editor Nilesh Mohite
Riddhi Tuljapurkar
Cover Work
Technical Editor Nilesh Mohite
Vivek Pala

Copy Editor
Tani Kothari
 About the Authors

Fábio M. Soares holds a master's degree in applied computing from UFPA and
is currently a PhD candidate at the same university. He has been designing neural
network solutions since 2004 and has developed applications with this technique in
several fields, ranging from telecommunications to chemistry process modeling, and
his research topics cover supervised learning for data-driven modeling.
He is also self-employed, offering services such as IT infrastructure management as
well as database administration to a number of small- and medium-sized companies
in northern Brazil. In the past, he has worked for big companies such as Albras, one
of the most important aluminium smelters in the world, and Eletronorte, a great
power supplier in Brazil. He also has experience as a lecturer, having worked at the
Federal Rural University of Amazon and as a Faculty of Castanhal, both in the state
of Pará, teaching subjects involving programming and artificial intelligence.
He has published a number of works, many of them available in English, all
including the topics of artificial intelligence applied to some problem. His
publications include conference proceedings, such as the TMS (The Minerals Metals
and Materials Society), Light Metals and the Intelligent Data Engineering and
Automated Learning. He has also has published two book chapters for Intech.

I would like to give a special acknowledgement to God for having

given me the opportunity to get access to rich knowledge on this
theme, which I simply love doing research on. Special thanks to my
family, my father, Josafá, and mother, Maria Alice (in memoriam),
who would be very proud of me for this book, and also my brother,
Flávio, my aunt, Maria Irenice, as well as all my relatives who always
supported me in some way during my studies. I would also like to
thank the support of my advisor, Prof. Roberto Limão. I am very
grateful to him for having invited me to work with him on many
projects regarding artificial intelligence and neural networks. Also,
special thanks to my partners and former partners from Exodus
Sistemas, who have helped me in my challenges in programming and
IT infrastructure. Finally, I'd like to thank my friend Alan Souza, who
wrote this book with me, for having extended to me this authorship.
Alan M.F. Souza is computer engineer from Instituto de Estudos Superiores
da Amazônia (IESAM). He holds a post-graduate degree in project management
software and a master's degree in industrial processes (applied computing)
from Universidade Federal do Pará (UFPA). He has been working with neural
networks since 2009 and has worked with IT Brazilian companies developing in
Java, PHP, SQL, and other programming languages since 2006. He is passionate
about programming and computational intelligence. Currently, he is a professor at
Universidade da Amazônia (UNAMA) and a PhD candidate at UFPA.

Since I was a kid, I thought about writing a book. So, this book is a
dream come true and the result of hard work. I'd like to thank God
for giving me this opportunity. I'd also like to thank my father, Célio,
my mother, Socorro, my sister, Alyne, and my amazing wife, Tayná,
for understanding my absences and worries at various moments. I
am grateful to all the members of my family and friends for always
supporting me in difficult times and wishing for my success. I'd like
to thank all the professors who passed through my life, especially
Prof. Roberto Limão for introducing me the very first neural network
concept. I must register my gratitude to Fábio Soares for this great
partnership and friendship. Finally, I must appreciate the tireless
team at Packt Publishing for the invitation and for helping us in the
production process as a whole.
 About the Reviewer

Saeed Afzal, also known as Smac Afzal, is a professional software engineer and
technology enthusiast based in Pakistan. He specializes in solution architecture and
the implementation of scalable high-performance applications.

He is passionate about providing automation solutions for different business needs

on the Web. His current research and work includes the futuristic implementation of
a next-generation web development framework, which reduces development time
and cost and delivers productive websites with many necessary and killer features by
default. He is hopeful of launching his upcoming technology in 2016.

He has also worked on the book Cloud Bees Development by Packt Publishing.

You can found out more about his skills and experience at http://sirsmac.com.
He can be contacted at [email protected].

I would like to thank the Allah Almighty, my parents, and my wife,

Dr. H. Zara Saeed, for all their encouragement.
 www.PacktPub.com

Support files, eBooks, discount offers, and more

For support files and downloads related to your book, please visit www.PacktPub.com.

Did you know that Packt offers eBook versions of every book published, with PDF
and ePub files available? You can upgrade to the eBook version at www.PacktPub.
com and as a print book customer, you are entitled to a discount on the eBook copy.
Get in touch with us at [email protected] for more details.

At www.PacktPub.com, you can also read a collection of free technical articles, sign
up for a range of free newsletters and receive exclusive discounts and offers on Packt
books and eBooks.
TM

https://www2.packtpub.com/books/subscription/packtlib

Do you need instant solutions to your IT questions? PacktLib is Packt's online digital
book library. Here, you can search, access, and read Packt's entire library of books.

Why subscribe?
• Fully searchable across every book published by Packt
• Copy and paste, print, and bookmark content
• On demand and accessible via a web browser

Free access for Packt account holders

If you have an account with Packt at www.PacktPub.com, you can use this to access
PacktLib today and view 9 entirely free books. Simply use your login credentials for
immediate access.
 Table of Contents
Preface vii
Chapter 1: Getting Started with Neural Networks 1
Discovering neural networks 2
Why artificial neural network? 2
How neural networks are arranged 4
The very basic element – artificial neuron 5
Giving life to neurons – activation function 5
The fundamental values – weights 6
An important parameter – bias 7
The parts forming the whole – layers 7
Learning about neural network architectures 8
Monolayer networks 8
Multilayer networks 9
Feedforward networks 9
Feedback networks 10
From ignorance to knowledge – learning process 10
Let the implementations begin! Neural networks in practice 11
Summary 18
Chapter 2: How Neural Networks Learn 19
Learning ability in neural networks 19
How learning helps to solve problems 20
Learning paradigms 20
Supervised learning 20
Unsupervised learning 21
Systematic structuring – learning algorithm 22
Two stages of learning – training and testing 23
The details – learning parameters 24
Error measurement and cost function 25

[i]
Table of Contents

Examples of learning algorithms 26

Perceptron 26
Delta rule 27
Coding of the neural network learning 27
Learning parameter implementation 27
Learning procedure 29
Class definitions 30
Two practical examples 37
Perceptron (warning system) 37
ADALINE (traffic forecast) 41
Summary 46
Chapter 3: Handling Perceptrons 47
Studying the perceptron neural network 48
Applications and limitations of perceptrons 48
Linear separation 48
Classical XOR case 50
Popular multilayer perceptrons (MLPs) 52
MLP properties 52
MLP weights 53
Recurrent MLP 54
MLP structure in an OOP paradigm 55
Interesting MLP applications 56
Classification in MLPs 56
Regression in MLPs 58
Learning process in MLPs 60
Simple and very powerful learning algorithm – Backpropagation 61
Elaborate and potent learning algorithm – Levenberg–Marquardt 63
Hands-on MLP implementation! 65
Backpropagation in action 68
Exploring the code 68
Levenberg–Marquardt implementation 72
Practical application – types of university enrolments 75
Summary 78
Chapter 4: Self-Organizing Maps 79
Neural networks' unsupervised way of learning 80
Some unsupervised learning algorithms 80
Competitive learning or winner takes all 82

[ ii ]
 Table of Contents

Kohonen self-organizing maps (SOMs) 84

One-Dimensional SOM 85
Two-Dimensional SOM 86
Step-by-step of SOM learning 88
How to use SOMs 89
Coding of the Kohonen algorithm 90
Exploring the Kohonen class 92
Kohonen implementation (clustering animals) 95
Summary 98
Chapter 5: Forecasting Weather 99
Neural networks for prediction problems 100
No data, no neural net – selecting data 101
Knowing the problem – weather variables 102
Choosing input and output variables 103
Removing insignificant behaviors – Data filtering 103
Adjusting values – data preprocessing 104
Equalizing data – normalization 105
Java implementation for weather prediction 107
Plotting charts 107
Handling data files 108
Building a neural network for weather prediction 109
Empirical design of neural networks 112
Choosing training and test datasets 112
Designing experiments 113
Results and simulations 113
Summary 116
Chapter 6: Classifying Disease Diagnosis 117
What are classification problems, and how can neural networks
be applied to them? 118
A special type of activation function – Logistic regression 119
Multiple classes versus binary classes 120
Comparing the expected versus produced results – the
confusion matrix 121
Classification measures – sensitivity and specificity 122
Applying neural networks for classification 123
Disease diagnosis with neural networks 126
Using ANN to diagnose breast cancer 126
Applying NN for an early diagnosis of diabetes 131
Summary 134

[ iii ]
Table of Contents

Chapter 7: Clustering Customer Profiles 135

Clustering task 136
Cluster analysis 137
Cluster evaluation and validation 138
External validation 138
Applied unsupervised learning 139
Neural network of radial basis functions 139
Kohonen neural network 140
Types of data 141
Customer profiling 142
Preprocessing data 142
Implementation in Java 143
Card credit analysis for customer profiling 143
Summary 149
Chapter 8: Pattern Recognition (OCR Case) 151
What is pattern recognition all about? 152
Definition of classes among tons of data 152
What if the undefined classes are undefined? 153
External validation 154
How to apply neural networks in pattern recognition 154
Preprocessing the data 155
The OCR problem 156
Simplifying the task – digit recognition 156
Approach to digit representation 157
Let the coding begin! 157
Generating data 158
Building the neural network 160
Testing and redesigning – trial and error 161
Results 163
Summary 166
Chapter 9: Neural Network Optimization and Adaptation 167
Common issues in neural network implementations 168
Input selection 168
Data correlation 169
Dimensionality reduction 170
Data filtering 171
Structure selection 171

[ iv ]
 Table of Contents

Online retraining 172

Stochastic online learning 174
Implementation 174
Application 176
Adaptive neural networks 179
Adaptive resonance theory 179
Implementation 180
Summary 182
Appendix A: Setting up the NetBeans Environment 183
Download and install NetBeans 183
Setting up the NetBeans environment 188
Importing a project 191
Programming and running code with NetBeans 194
Debugging with NetBeans 195
Appendix B: Setting Up the Eclipse Environment 199
Download and install Eclipse 199
Setting up the Eclipse environment 204
Importing a project 205
Programming and running code with the Eclipse IDE 210
Debugging with the Eclipse IDE 211
Appendix C: References 213
Chapter 1 – Getting Started with Neural Networks 213
Chapter 2 – How Neural Networks Learn 213
Chapter 3 – Working with Perceptrons 213
Chapter 4 – Self-Organizing Maps 214
Chapter 5 – Forecasting the Weather 214
Chapter 6 – Disease Diagnosis 214
Chapter 7 – Clustering Customer Profiles 215
Chapter 8 – Pattern Recognition (the OCR Case) 215
Chapter 9 – Neural Network Optimization and Adaptation 215
Index 217

[v]
 Preface
The life of a programmer can be described as a continual never-ending learning
pathway. A programmer always faces challenges regarding new technology or new
approaches. Generally, during our lives, although we become used to repeated
things, we are always subjected to learn something new. The process of learning is
one of the most interesting topics in science, and there are a number of attempts to
describe or reproduce the human learning process.

The writing of this book was guided by the challenge of facing new content and
then mastering it. While the name neural networks may appear strange or even give
an idea that this book is about neurology, we strived to simplify these nuances by
focusing on your reasons for deciding to purchase this book. We intended to build
a framework that shows you that neural networks are actually simple and easy to
understand, and absolutely no prior knowledge on this topic is required to fully
understand the concepts we present here.

So, we encourage you to explore the content of this book to the fullest, beholding
the power of neural networks when confronting big problems but always with the
point of view of a beginner. Every concept addressed in this book is explained in easy
language, and also with a technical background. Our mission in this book is to give
you an insight into intelligent applications that can be written using a simple language.

Finally, we would like to thank all those who directly or indirectly have contributed
to this book and supported us from the very beginning, right from the Federal
University of Pará, which is the university that we graduated from, to the data and
component providers INMET (Brazilian Institute of Meteorology), Proben1, and
JFreeCharts. We want to give special thanks to our advisor Prof. Roberto Limão, who
introduced us to the subject of neural networks and coauthored many papers with
us in this field. We also acknowledge the work performed by several authors cited
in the references, which gave us a broader vision on neural networks and insights on
how to adapt them to the Java language in a didactic way.

[ vii ]
Exploring the Variety of Random
Documents with Different Content
 FOUNDATIONS UNDER RECIPROCATING ENGINES 153 (h3
is the distance between the center of gravity of the mat and the
foundation contact area). For the upper part of the foundation,
located above the mat, the moment of inertia, from an analogous
formula, is Q OK /04 = -j~ (4.82 + 1.02) + 3.25 X LO2 = 9.8 tons X
m X sec2 The total moment of inertia of the mass of the whole
system with respect to this axis is 4 Wo = £ hi = 7.3 +3.1 +6.1 +
9.8 = 26.3 tons X m X sec2 i = l The moment of inertia of the whole
system with respect to the axis passing through the center of gravity
of the whole system perpendicular to the plane of vibrations is I = /0
_ mh* = 26.3 - 6.91 X 1.092 = 18.2 tons X m X sec2 since h = z0 =
1.09 m. The ratio between the moments of inertia is 7 = 18.2/26.3
= 0.69 7. Computation of Amplitudes of Forced Vibrations of a
Foundation Accompanied Simultaneously by Sliding and Rocking.
The limiting natural frequency of rocking vibrations of the
foundation, according to Eq. (III-2-6), is w _ 10 X 103 x 54 - 67.5 X
LOO _ ^ x ^ ^ The limiting frequency of vibrations in shear, from
Eq. (III-3-2), is . . 2.5 X 103 X 18 c _ ^ 1Aa _2 fnx2 = g-jn: = 6.5 X
103 sec 2 The frequency equation for the foundation [Eq. (III-4-8)]
is , 4 _ (20.5 + 6.5) X 103 , 2 , 20.5 X 6.5 _ Jn 0.69 U i~ 0.69 " or
fn* - 39.2 X 103/n2 + 193.0 X 106 = 0 By solving this equation we
find the natural frequencies of vibrations of the system : /m2 = 33.4
X 103 sec"2 /n22 = 5.8 X 103 sec"2 We compute the coefficient
A(co2): A(C02) = mlUnl* - "2)(/n22 - CO2) = 6.91 X 18.2(33.4 - 2.5)
(5.8 - 2.5) X 106 = 13.8 X 109
 154 DYNAMICS OF BASES AND FOUNDATIONS From Eqs.
(III-4-12) we determine the amplitudes of sliding shear and rotation
of the foundation. The amplitude of sliding shear of the center of
gravity of the whole system is , 2.5 X 103 X 18 X 1.09 nni«N/in-»
nm« A;v = — 0 4.6 = 0.016 X 10 3 m = 0.016 mm The amplitude of
rotation is 2.5 X 103 X 18 - 6.91 X 2.5 X 103 , _ _ nnn ^ in_. .. A
 FOUNDATIONS UNDER RECIPROCATING ENGINES 155 2.
Selection of a Design Diagram for the Foundation. The dimensions of
the foundation are to be selected according to design considerations
based on the requirements of the plant management. Figure IV-8
gives the design diagram selected on the basis of these
considerations. Concrete type 100 f is to be employed for the
foundation. 3. Centering of the Foundation Area in Contact with Soil
and Determination of Pressure on the Soil. Let us determine the
coordinates Xo, yo, and z0 of -2210-960-> Section along longitudinal
axis of the foundation -1950J *-750>H-750 r -1.20 V. . . ■1950
±0.00 _5Z '///////////'/s/// •3000 '///////St -30006000_1 T -2.00
Section along axis of main shaft Fig. IV-8. Design diagram of
foundation, example 2. the center of gravity of the whole system
(the foundation and engine) with respect to the axes shown in Fig.
IV-8: "SrriiXi m yo Xnnyi liiriiZi where m» = masses of separate
elements of system Xij y%, Zi = coordinates of centers of gravity of
these elements with respect to x, y, z axes m = mass of complete
system Separate elements of the foundation are marked in Fig. IV-8
by the numbers 1, 2, and so on. The foundation should be divided
into elements of such shape that the data on magnitudes of masses
and coordinates of centers of gravity of separate elements may f See
footnote in Art. IV-l-c, p. 132.
 156 DYNAMICS OF BASES AND FOUNDATIONS be used
later, when the moment of inertia of the mass of the whole system
will be computed. The data for the computation of coordinates of
the center of gravity of the system are given in Table IV-2. Masses
corresponding to cavities in the foundation are shown with minus
signs. From the data of Table IV-2 we find the coordinates of the
center of gravity of the system : 89.14 64.52 28.55 Xo = oi g = 4-16
m y° ~ 01 E == 3-05 m z0 = = 1.33 m The displacement of the
center of gravity of the mass of the system with respect to the
center of the foundation area in contact with soil is: In the direction
of the x axis: 4.16 - 4.00 = 0.16 m In the direction of the y axis:
3.05 - 3.00 = 0.05 m The relative magnitude of the eccentricity in
the direction of the x axis is n i (\ -—- 100 = 2 per cent which is less
than 5 per cent. The eccentricity in the direction of the y axis is even
smaller. Since the eccentricity in the mass distribution is small, its
influence on the amplitudes of forced vibrations will be insignificant.
Therefore we neglect hereafter the eccentricity and consider that the
center of gravity of the mass of the system and the centroid of the
foundation contact area are located on the same vertical line. The
pressure on the soil imposed by the static load is assumed to be
uniformly distributed over the foundation contact area and equals
mg 21.5 X 9.81 . . , , . n .. . , 2 pBt = — p = j£— — = 4.4 tons/m2
= 0.44 kg/cm2 Thus the static pressure on the soil is considerably
smaller than the permissible pressure. 4. Computation of Amplitudes
of Forced Vibrations of the Foundation. Since the horizontal
component of the unbalanced inertial forces of the engine in the
direction perpendicular to the motion of the piston is zero, and since
the vertical component of the above forces is insignificant, we
compute the amplitudes of forced vibrations only for foundation
vibrations caused by the horizontal component of unbalanced inertial
forces in the direction of piston motion (the system will be sub
jected to vibrations in the xz plane). We also neglect the action of
exciting moments tending to produce rocking vibrations of the
foundation. We begin by establishing the data needed for the
computation of amplitudes of foundation vibrations. The frequency
of machine rotation equals to = 0.105V = 0.105 X 167 = 17.3 sec"1
w2 = 300 see-2
The text on this page is estimated to be only 6.84%
accurate

^ = N 00 to CO •* CO N CO CT> + OS t^. O O CO o o o *.
W(D OO H N O CN o | | 1—1 H i =3 >> betwee gravit nt and center
ity, m N N H 00 C N 00 IfS IO 00 O 00 o o .-I .-i © © o o o m «w 5 0
> SogoJ s s? * a Sb j -2 « a «r OS tJH CD •*}< CD H CO ■* CD ■*
IO «£ d •« o o IS a> o o Q o © i-h o o o i-i cn o iO '. ; oo h h d d
CD © Moment i mass of el respect to through gravity i tons X e • • •
N * || 00 iTa «M N !D t(i O) H iO O) iO i O o n ■ | 49 s (O ^tH (ON
H H ^ 1 1 CN * g S i WN^hOhN CN -i-3 ci »<0 H 00 M O) O tO g£ -
x 3) I © ^ ID O) CO N CN Tf i % * 1 as »"5 HM II CO •3 I WO N 00
CONN ■* H O O »0 O rH CN CO •"J o H © t» ^ N >C || 00 es of
center of element spect to axes, m O O >0 >C OS O >0 *4 » Ol IN
N H Tf N N NO HN H H t£ 00 O O O O bS 00 h o o o o o "^ ij £ o h a
N . o CO N 00N O N IO 03 43 oa 03 .» CN tJH CO CN rH CT> OS IO
|x s CN -H »C CO rH © © -1 1 1 CN § O *° • • o o b- o o N • • UJ
UJ CO N >0 la §5 • e • 'ohOhh • • O © i-O © "O •s a a • O "# Oi IO
CN 2 a e • '(0 10HHH a § Q^ • O O CD O to ■ O CO CO "* CN • • oo
r- CO CN CN _, ^ l-l " a g d « * a o o 03 03 4) * 4) 43 . Elem of sys
(engin d 3 Z> aCN ' 1 id 1 '3 O 157
 158 DYNAMICS OF BASES AND FOUNDATIONS The
distance from the axis of the master shaft of the engine to the
common center of gravity of the mass of the system is hi = 1.53 m.
The exciting moment of the engine is then M = PJbx = 12.8 X 1.53
= 19.6 tons X m The moment of inertia of the foundation contact
area with respect to the axis passingthrough its center of gravity
perpendicular to the plane of vibrations equals - 6.00 X 8.003 ._. , /
= — = 256 m4 The weight of the whole system is W = mg - 21.5 X
9.81 = 211 tons The moment of inertia of the mass of the whole
system with respect to the axis passing through the common center
of gravity perpendicular to the plane of vibrations equals Wm =
K22rrii(aXi* 4- azi*) 4- m2(z0»2 + z0i2) = 89.69 + 12.09 = 101.78
^ 102 tons X m X sec2 The moment of inertia of the mass of the
whole system with respect to the axis passing through the centroid
of the foundation contact area perpendicular to the plane of
vibrations equals Wo + Im = mh2 = 102 + 21.5 X 1.332 = 140 tons
X m X sec2 The ratio between the moments of inertia of the masses
is 7 = 102i4o = 0.73 The limit value of natural frequency of rocking
vibrations of the foundation is determined from Eq. (III-2-6): M _ 8
X 103 x 2se_ L33 x 21.1 _ ^ 6 x 10S sec_2 The limit value of the
natural frequency of sliding shear vibrations, from Eq. (III-3-2), is =
2 X 10' X 48 = 4 4(, x 1Q3 gec_2 21.5 We set up the frequency
equation of the foundation according to Eq. (III-4-8): 4 _ 14.6 X 103
+ 4.46 X 103 2 14.6 X 103 X 4.46 X 103 n fn 0.73 /n + 0.73 fn4 -
26.0 X 103/,,2 + 89.0 X 106 = 0 Solving this equation, /m.22 =
[13.0 ± V(13.02 - 89.0)]103 Hence /nl.22 = (13.0 ± 8.9)103 Thus
the natural frequencies of the foundation will be fnl = 21.9 X 103
sec"2 /„22 = 4.1 X 103 sec"2
 FOUNDATIONS UNDER RECIPROCATING ENGINES 159 We
compute the coefficient: A(co2) =mWm(fnli - "2)(/„22 - «») = 21.5
X 102(21.9 - 0.30) (4.1 - 0.30) X 106 = 18.0 X 1010 We then
compute the amplitudes of forced vibrations induced by the
horizontal force Px and by the moment M = PJir, according to Eqs.
(III-4-11) and (III-4-12) the horizontal displacement of the common
center of gravity of the foundation and the engine is (8 X 103 X 256
- 211 X 1.33 + 2 X 103 X 48 X 1.33 - 102 X 17.32)12.8 Ax 18.0 X
1010 , 2 X 103 X 48 X 1.33 X 19.6 ni-vin_3 n ,+ 18.0 X 1010 —
=0.17 mm From the same equations we find the amplitude of
rocking vibrations of the foundation about the horizontal axis passing
through the center of gravity of the foundation perpendicular to the
plane of vibrations: (2 X 103 X 48 X 1.33 X 12.8 + (2 X 103 X 48 -
21.5 X 17.32)19.6 v ~ 18.0 X 1010 = 0.019 X 10-3 radians Thus the
amplitude of forced vibrations of the upper edge of the foundation
equals A = 0.17 X 10~3 + 1.04 X 0.019 X 10~3 = 0.19 10"3 m <
0.2 mm The design value of the amplitude of vibrations does not
exceed the permissible value; hence the dimensions of the
foundation are selected correctly. The foregoing computations show
that vibrations of the foundation are produced mainly by its
horizontal displacement in the direction of the action of the
horizontal component of the disturbing force of the compressor. This
is explained by the fact that the dimensions of the foundation in the
direction of the action of this force is large in comparison with the
height of the foundation. Therefore rocking results only in small
dynamic displacements. Hence the rocking vibrations of a foundation
may be neglected when computing the amplitude of forced
vibrations if the foundation is elongated in the direction of action of
the horizontal exciting force; in this case the vibrations of the
foundation may be considered to be vibrations of sliding shear. This
assumption greatly simplifies dynamic computations. However, this
simplification of computations should be very cautiously applied. For
example, if such a simplification were made in the case of the
foundation under consideration, then from Eq. (III-3-3) we would
obtain for the amplitude of horizontal displacements of the
foundation = 0.14 X 10-3m = 0.14 mm 21.5(4.46 - 0.30) X 103 The
computed amplitude is 26 per cent smaller than that obtained by
means of the foregoing computations (0.19 mm). This cannot be
admitted as a good approximation. The results of computations will
be more accurate if we add to the amplitude of
 160 DYNAMICS OF BASES AND FOUNDATIONS vibrations of
sliding shear a displacement produced by rocking vibrations of the
foundation, computed from the formula where A? is the amplitude of
rocking vibrations when no shear is present, determined from Eq.
(III-2-7), and h is the full height of the foundation. Assuming that
the horizontal exciting force acts at height hi from the base of the
foundation, and assuming that, in Eq. (III-2-7), M — Pxhx, we
obtain , Pxhhx Ax — Wo{jn2 ~ CO2) For the foundation under
consideration, h = 2.0 m and hi = 2.9 m; consequently, A" " 14oi?
£X-^W)2X10» = °-°38 X1°"'m " °-°4 mm Thus the total amplitude
of the horizontal displacement of the foundation will equal A = Ax +
Ax
 FOUNDATIONS UNDER RECIPROCATING ENGINES 161 by
the second method, even if this leads to some increase in vertical
vibrations. Therefore, if such an engine was counterbalanced by
means of the first method but impermissible horizontal vibrations
were observed after the construction of the foundation, then
counterbalancing by means of the second method (i.e., by changing
the character of counterbalancing) may be recommended as one of
the simplest measures to decrease these vibrations. In cases in
which vertical vibrations of an impermissible amplitude are present
in systems with horizontal motors, the second method is unsuitable,
and the first method should be applied. Similarly, for a vertical
motor, the method of counterbalancing selected depends on the type
of foundation vibrations — vertical, horizontal, or rocking. The
installation of counterweights for balancing a motor does not require
dismantling or prolonged interruption of operation. The interruption
is only for the time needed to attach the counterweight to the sides
of the crank. b. Chemical Stabilization of Soils. If a foundation rests
on sandy soil, then, in order to decrease vibrations, chemical or
cement stabilization of the soil under the foundation may be used.
Such soil stabilization will result in an increase in the rigidity of the
base and consequently in an increase in the natural frequencies of
the foundation. Therefore this method is very effective when natural
frequencies of the foundation on a nonstabilized soil are higher than
the operational frequency of the engine — which is usually the case.
An increase in rigidity will increase still further the difference
between the frequency of natural vibrations and the frequency of the
engine; consequently the amplitudes of foundation vibrations will
decrease. When a foundation resting on a natural soil has natural
frequencies smaller than the operational frequency of the engine,
then soil stabilization may cause an increase in the amplitudes of
vibrations. This may be undesirable if a soil is stabilized to such a
degree that frequencies of natural vibrations of the foundation
merely approach the operational frequency. But if a soil is thoroughly
stabilized and natural frequencies of the foundation became much
higher than the operational frequency of the engine, then such soil
stabilization may result in a considerable decrease in amplitudes of
vibrations. Chemical and cement stabilization of soils is economically
advantageous, since its costs are low in comparison, for example,
with structural measures. The principal advantage of this method lies
in the fact that it can be applied without a prolonged interruption in
the work of the engine. The interruption is only for the period of
direct work connected with soil stabilization and then for 2 to 3 days
more. Thus the over-all result is that the engine will be inactive only
for a few days.
 162 DYNAMICS OF BASES AND FOUNDATIONS The limits of
the stabilized zones of soil and their shape are determined by the
character of the vibrations. If, for example, a foundation is subjected
mainly to rocking vibrations about an axis passing through the
centroid of the base contact area, then it suffices to stabilize the soil
near the foundation edges, perpendicular to the plane of vibrations,
and it is not necessary to stabilize the soil under the entire
foundation. The depth of the stabilized zone should be no less than
1 to 2 m. This method of decreasing vibrations was applied at one of
the Soviet plants when it became necessary to decrease the
amplitudes of vibrations of an operating horizontal compressor
without a long interruption in its work. Soil was stabilized to a depth
of about 1.0 m; the zone extended horizontally 30 cm beyond the
foundation edges. The results of foundation vibration measurements
before and after stabilization showed that the amplitudes of
vibrations, on the average, decreased by 50 per cent. The work of
the compressor was stopped only for the period of injection of the
silicates; the engine was set in motion immediately after silicatization
was completed. It can therefore be assumed that when the
compressor renewed its motion, the stabilized soil had not as yet
formed a sufficiently rigid base, and it is possible that foundation
vibrations acted unfavorably on the stabilized zone of soil, which had
not fully hardened. c. Structural Measures. The use of structural
measures for decreasing foundation vibrations often requires a long
interruption in the engine's operation and considerable expense of
funds and materials. Therefore the use of this method may be
suggested only in cases in which for some reason no other methods
may be applied. At the same time, it should be noted that the
correct change in foundation design may prove very effective in
decreasing the amplitude of vibrations. Structural measures are
applied with the purpose of changing the natural frequencies of a
foundation in such a way as to achieve the largest possible
difference between them and the operational frequency of the
engine. The choice of structural measure depends on the nature of
the vibrations and the interrelationships between the frequencies of
natural and forced vibrations. The operational frequencies of
reciprocating engines are usually lower than the fundamental
frequencies of foundations; therefore most of the structural
measures are directed towards increasing still further the natural
frequencies of the foundation. This is achieved by increasing the
foundation contact area and its moments of inertia, as well as by
increasing the rigidity of its base by means of piles. In addition, it is
possible to increase the foundation mass without inducing changes
in the frequency of foundation vibrations. This results in a decrease
in the amplitudes of vertical vibrations. When check calculations of
the natural frequencies of a vibrating foundation show that they are
lower than the operational frequencies of
 FOUNDATIONS UNDER RECIPROCATING ENGINES 163 the
engine, an enlargement of the foundation contact area or an
increase in the soil rigidity not only may not decrease the amplitudes
of vibrations, but may even increase them. In this case, it is better
to decrease still more the natural frequency of the foundation. This
may be achieved by enlarging the foundation mass without an
increase in its area in contact with soil. The selection of particular
structural measures depends on local conditions. For example, if a
vibrating foundation lies close to another foundation, it can be
attached to the latter. As an illustration we will Fig. IV-9. Machine
foundation which developed horizontal vibrations of high amplitude.
describe here the structural measures which were applied to a
foundation under a horizontal compressor in order to decrease its
vibrations. The horizontal component of the exciting forces induced
by the whole system was 30 tons. The foundation rested on a
medium-grained sand with clay laminae. The foundation consisted of
a block about 4.6 m high, with a base area 7 by 8 m2, placed on 55
situ-cast piles. The length of the piles was about 8 m. Figure IV-9
shows a cross section of the foundation. Horizontal vibrations of
extremely large amplitude (around 0.9 mm) were observed while the
engine was in operation. At the same time, there occurred
settlement of the basement of an adjoining structure, under which
no piles were provided. On the side nearest the foundation under
discussion, settlement of the basement reached 70 mm. It appears
that this considerable settlement of the basement was caused by
vibrations transmitted from the foundation. These vibrations
furthered the loosen 
 164 DYNAMICS OF BASES AND FOUNDATIONS ing of the
soil and the carrying away of soil particles from beneath the
foundation and basement by ground water. When reinforcement of
the foundation was started, it was found that soil under it was
washed out or had subsided to a depth of about 0.5 m. No damage
was found in the foundation block. The ground-water level was
approximately 1.5 m above the level of the foundation base.
Reinforcement of the foundation was undertaken with the purpose
of decreasing its vibrations. After the ground-water level had been
artiNew piles Old piles Fig. IV-10. Reinforcement of foundation
shown in Fig. IV-9. ficially lowered, soil was removed from beneath
the foundation to a depth of 0.75 m. Thus an excavation of a total
depth of about 1.25 m was formed. This excavation was filled with
concrete. The foundation area in contact with the soil was extended
on all sides and 33 new situcast piles were installed. Figure IV-10
illustrates the measures recommended for reinforcement.
Reinforcement of the new part of the foundation provided a good
connection with the old part. To avoid settlement of the footings
under the building walls due to excavation of the soil, sheetpiling
and chemical stabilization of soil were used beneath the footings
under the walls. This reinforcement of the foundation was very
effective. The amplitudes of foundation vibrations decreased from
0.9 to 0.05 mm, i.e., 18 times. This decrease in amplitudes was
caused by considerable increase in the natural frequencies of
foundation vibrations due to an increase in the
 ^ i ? FOUNDATIONS UNDER RECIPROCATING ENGINES
165 foundation area in contact with the soil, as well as to an
increase in the moment of inertia of the contact area. In addition, a
considerable effect was produced by the extra foundation mass and
the increase in the rigidity of the base due to the installation of
supplementary piles. To facilitate the application of various structural
measures for decreasing foundation vibrations, it is recommended in
doubtful cases to leave projecting reinforcement which may be used,
if necessary, for the attachment of an additional mass to the
foundation or for the extension of its area in contact with soil. These
measures, of course, should be applied only after recognizing the
fact that the foundation is undergoing 1 px s'mfl* vibrations of an
impermissible ] ( ( ^A magnitude. The use of special slabs, first
proposed by Professor N. P. Pavliuk and Engineer A. D. Kondin^ may
^ iyn Analysis of the effect of ^ be considered as a structural meas-
taching a slab A to an engine foundation, ure to decrease foundation
vibrations. By means of these slabs, it is possible in some cases to
decrease the amplitudes of rocking and horizontal vibrations of
foundations. Let us assume (Fig. IV-11) that slab A, resting on soil,
is attached to a foundation undergoing rocking vibrations around the
axis passing through the centroid of the foundation area in contact
with soil. Let us set up the equation of forced rocking vibrations of
the foundation. The following symbols will be used: Wo = moment
of inertia of foundation mass and of mass of engine with respect to
axis of vibrations / = moment of inertia of foundation area in contact
with soil, with respect to same axis PT sin ut = magnitude of
horizontal exciting force induced by engine and transmitted to
foundation, where co = frequency of engine rotation H = distance
between line of action of exciting force and foundation contact area
h = distance between center of mass of foundation and engine, and
foundation contact area mi = mass of attached slab A i = contact
area of attached slab Hi = distance between place of connection of
foundation with attached slab, and foundation contact area W =
foundation weight cv, cr = coefficients of elastic nonuniform
compression, shear of soil.
The text on this page is estimated to be only 29.22%
accurate

166 DYNAMICS OF BASES AND FOUNDATIONS The

differential equation of forced vibrations of the foundation together
with the attached slab will be as follows: (Wo + miflV)^ + (c+I - Wh
+ H^CrAJv = PTH sin ut (IV-5-1) From this we obtain in the usual
way the expression for the natural frequency of rocking vibrations of
the foundation with attached slab : a _ Cyl - Wh + H^CrAx Jn U2
(IV-5-4) , _ cj - Wh or since fnip2 = ^ we substitute into the left-
hand part of (IV-5-4) the expression for fntpi2 from Eq. (IV-5-2);
then, neglecting the term containing Wh because of its smallness,
we obtain IFo + m^ - Wl (1V'5_5) Hence CrA_i > c^I ■ mi W 0 or
fnxl* > /n,2 (IV-5-6) Thus if one selects the attached slab so that
the frequency of its natural vibrations of pure shear equals the
frequency of rocking vibrations of the foundation, then the attached
slab will have no effect on the magnitude of the frequency of natural
vibrations of the foundation. Besides, the amplitude of forced
vibrations of the foundation will decrease according to the ratio A9l 1
Av 1 + 5 (IV-5-7) * ^i#i2 where 8 = -^= — Wo
 FOUNDATIONS UNDER RECIPROCATING ENGINES 167
Usually in foundations under engines, fn
 168 DYNAMICS OF BASES AND FOUNDATIONS freedom
may be directly applied when considering the forced vibrations of the
system. For example, for the amplitudes A\ and A2 of forced
vibrations of the foundation and the attached masses, respectively,
we have: A /222 ~ CO2 /r-2 S&) P (IV"5"9) Also, fna* = SL m2
where fna = natural frequency of vibrations of attached masses Co
= coefficient of elastic rigidity of tie between these masses and
foundation p = magnitude of exciting force per unit of foundation
mass A(co)2 = Unl2 ~ C02)(/n22 - CO2) (IV-5-10) where /„i and
fn% are natural fundamental frequencies of the foundation with
dampers. It follows from the first of Eqs. (IV-5-9) that the amplitude
of foundation vibrations becomes zero when U2=fnJ (IV-5-11) i.e.,
when the frequency of natural vibrations of the attached masses
equals the frequency of the exciting force. In order to determine the
amplitude of vibrations of the damper masses, we substitute /22 into
the right-hand part of Eq. (IV-5-10) in place of co; then mi
Substituting this expression for A(co2) into the second of Eqs. (IV-5-
9), we obtain the following expression for the amplitude of vibrations
of the damper: Since v = — — it follows that A2 = ^ (IV-5-12) c2
 FOUNDATIONS UNDER RECIPROCATING ENGINES 169
Thus, the amplitude of vibrations of the damper equals its static
deflection as produced by a force of magnitude equal to the
maximum value of the exciting force Pit). Equations (IV-5-11) and
(IV-5-12) determine the selection of the damper. It should be noted
that neither the frequencies of the damper nor the amplitudes of its
vibrations depend on the properties of the soil base or the mass of
the foundation. It follows from Eq. (IV-5-9) that theoretically it is
possible to damp vibrations of infinitely large foundations subjected
to the action of periodic exciting forces by attaching dampers to
these foundations, even dampers with small masses. However, the
smaller the mass of the damper, the smaller should be the rigidity c2
of its elastic tie with the foundation, and consequently the larger will
be the amplitude of its vibrations. At values of c2 smaller than a
certain limit, amplitudes of vibrations of the damper may attain
magnitudes endangering its strength. Therefore the minimum value
of the damper mass is limited by permissible values of stresses in
the elastic tie between the damper and the foundation. It has
already been mentioned that when dampers are used, the
foundation has not one, but two natural frequencies of vibrations,
determined as roots of the equation /n4 ~ [fm2 + (1 + rifnaVn2 +
fntfna2 (IV-5-13) where /„z2 = — d = cuA mi The roots of Eq. (IV-5-
13) are: /nl,22 = liifnS +/na2(l + /i) ± Vlfnz2 +/n«2(l + m)2]2 ~
4/n22/na2) where fnz = frequency of natural vertical vibrations of
foundation fna = co = average operational machine rotation n =
m2/nii = ratio between masses of dampers and foundation mass
When a damper is installed, fni will be larger than both fnz and co,
and fn2 will be smaller than these frequencies. Besides, either /ni or
fn2 will lie close to u, and the other will be close to fnz. Let us
assume that fnz > co. Then the lower fundamental frequency /»2
will be close to co; the higher one, fnh will be close to/n*. If the
engine has varying angular frequency, then, with the installation of
the damper, the danger arises that one of the values of co will
coincide with fn2, i.e., that resonance will occur with the lower
frequency of the system "foundation and damper." In this case, the
amplitudes of foundation vibrations
 170 DYNAMICS OF BASES AND FOUNDATIONS will be
sharply increased and the masses attached to the foundation will
work not as dampers, but as intensifiers of the vibrations. In order
to avoid such an intensification of vibrations, ra2 should be selected
so that the maximum decrease in operational frequency of the
engine, as compared with the average value of this frequency, is
smaller than the difference between the average operational
frequency of the engine coav (equaling the natural frequency /22 of
the damper) and the lower frequency fn2 of the foundation. Hence,
the following condition should be satisfied when the damper sizes
are selected : Jn2 < Wmin" Or y2(,U +fnJ(l + /i) - VlfnS + fna2(l +
M)P = ±fnZ%a'2) < 0>min Solving this inequality for /jl and noting
that f 2 - M 2 we obtain /n22(wav2 — Wmin2) + OJmin4 V> Let us
assume that the nonuniformity in engine speed is as follows: COav
Hence o>min = wav(l — e) Substituting this expression for comin
into the right-hand part of the inequality obtained for ju, s ^ - (i - *y
f9 v where 0 = ^ > 1 Wav If 0 < /3 < 1, then comax = coav(l + e)
and we obtain: M > (1 t f ~ ^2 e(2 + e) (IV-5-14) (1 + e)2 The
inequalities obtained show that the selection of a proper
interrelationship between the damper mass and the foundation mass
depends not only on the values of the irregularity in the engine
speed, but also on the interrelationship between the natural
frequency of the foundation and the average operational speed of
the engine. If the frequency of natural vibrations is higher than the
operational frequency of the engine (i.e., /? > 1), then the damper
mass should be
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

textbookfull.com