Soft Computing Applications in Industry

383 Pages·2008·11.96 MB·English

by  B. V. Babu

BhanuPrasad(Ed.)
SoftComputingApplicationsinIndustry
Studiesin Fuzzinessand Soft Computing,Volume226
Editor-in-Chief
Prof.JanuszKacprzyk
SystemsResearchInstitute
PolishAcademyofSciences
ul.Newelska6
01-447Warsaw
Poland
E-mail:[email protected]
Furthervolumesofthisseriescanbefoundonourhomepage:springer.com
Vol.212.ElisabethRakus-Andersson Vol.220.HumbertoBustince,
FuzzyandRoughTechniquesinMedical FranciscoHerrera,JavierMontero(Eds.)
DiagnosisandMedication,2007 FuzzySetsandTheirExtensions:
ISBN978-3-540-49707-3 Representation,AggregationandModels,2007
ISBN978-3-540-73722-3
Vol.213.PeterLucas,JoséA.Gàmez,
AntonioSalmerón(Eds.) Vol.221.GlebBeliakov,TomasaCalvo,
AdvancesinProbabilisticGraphical AnaPradera
Models,2007 AggregationFunctions:AGuide
ISBN978-3-540-68994-2 forPractitioners,2007
ISBN978-3-540-73720-9
Vol.214.IrinaGeorgescu
FuzzyChoiceFunctions,2007 Vol.222.JamesJ.Buckley,
ISBN978-3-540-68997-3 LeonardJ.Jowers
MonteCarloMethodsinFuzzy
Vol.215.PaulP.Wang,DaRuan,
Optimization,2008
EtienneE.Kerre(Eds.)
ISBN978-3-540-76289-8
FuzzyLogic,2007
ISBN978-3-540-71257-2 Vol.223.OscarCastillo,PatriciaMelin
Vol.216.RudolfSeising Type-2FuzzyLogic:Theoryand
TheFuzzificationofSystems,2007 Applications,2008
ISBN978-3-540-71794-2 ISBN978-3-540-76283-6
Vol.217.MasoudNikravesh,JanuszKacprzyk, Vol.224.RafaelBello,RafaelFalcón,
LoftiA.Zadeh(Eds.) WitoldPedrycz,JanuszKacprzyk(Eds.)
ForgingNewFrontiers:Fuzzy ContributionstoFuzzyandRoughSets
PioneersI,2007 TheoriesandTheirApplications,2008
ISBN978-3-540-73181-8 ISBN978-3-540-76972-9
Vol.218.MasoudNikravesh,JanuszKacprzyk, Vol.225.TerryD.Clark,JenniferM.Larson,
LoftiA.Zadeh(Eds.) JohnN.Mordeson,JoshuaD.Potter,
ForgingNewFrontiers:Fuzzy MarkJ.Wierman
PioneersII,2007 ApplyingFuzzyMathematicstoFormal
ISBN978-3-540-73184-9 ModelsinComparativePolitics,2008
ISBN978-3-540-77460-0
Vol.219.RolandR.Yager,LipingLiu(Eds.)
ClassicWorksoftheDempster-ShaferTheory Vol.226.BhanuPrasad(Ed.)
ofBeliefFunctions,2007 SoftComputingApplicationsinIndustry,2008
ISBN978-3-540-25381-5 ISBN978-3-540-77464-8
Bhanu Prasad (Ed.)
Soft Computing
Applications in Industry
ABC
Editor
Prof.BhanuPrasad
FloridaA&MUniversity
DepartmentofComputerand
InformationSciences
Tallahassee,FL32307
USA
E-mail:[email protected]
AssociateEditors
JasonBlack
FloridaA&MUniversity,Tallahassee,FL,USA
BobbyGranville
FloridaA&MUniversity,Tallahassee,FL,USA
ZoranMajkic
UniversityofBelgrade,Belgrade,Serbia
YenumulaB.Reddy
GramblingStateUniversity,Grambling,LA,USA
ISBN978-3-540-77464-8 e-ISBN978-3-540-77465-5
DOI10.1007/978-3-540-77465-5
StudiesinFuzzinessandSoftComputing ISSN1434-9922
LibraryofCongressControlNumber:2007943077
(cid:2)c 2008Springer-VerlagBerlinHeidelberg
Thisworkissubject tocopyright. Allrightsarereserved, whetherthewholeorpart ofthematerial
isconcerned,specificallytherightsoftranslation,reprinting,reuseofillustrations,recitation,broad-
casting, reproduction on microfilm or in any other way, and storage in data banks. Duplication of
thispublicationorpartsthereofispermittedonlyundertheprovisionsoftheGermanCopyrightLaw
ofSeptember 9, 1965, initscurrent version, andpermission for use must always be obtained from
Springer.ViolationsareliabletoprosecutionundertheGermanCopyrightLaw.
Theuseofgeneraldescriptivenames,registerednames,trademarks,etc.inthispublicationdoesnot
imply, even in the absence of a specific statement, that such names are exempt from the relevant
protectivelawsandregulationsandthereforefreeforgeneraluse.
Typeset&CoverDesign:ScientificPublishingServicesPvt.Ltd.,Chennai,India.
Printedinacid-freepaper
987654321
springer.com
Contents
Optimization of Industrial Processes Using Improved and
Modified Differential Evolution
B.V. Babu, Rakesh Angira ......................................... 1
Noise-Robust Tool Condition Monitoring in Micro-milling
with Hidden Markov Models
K.P. Zhu, Y.S. Wong, G.S. Hong................................... 23
Dynamically Self-generated Fuzzy Neural Networks with
Industry Applications
Meng Joo Er, Yi Zhou............................................. 47
Kernel Estimators in Industrial Applications
Piotr Kulczycki ................................................... 69
Negative Selection Algorithm with Applications in Motor
Fault Detection
X.Z. Gao, S.J. Ovaska, X. Wang ................................... 93
Harmony Search Applications in Industry
Zong Woo Geem .................................................. 117
Soft Computing in Bioinformatics: Genomic and Proteomic
Applications
James Malone .................................................... 135
Evolutionary Regression and Neural Imputations of Missing
Values
Pawan Lingras, Ming Zhong, Satish Sharma.......................... 151
Glowworm Swarm Optimization Algorithm for Hazard
Sensing in Ubiquitous Environments Using Heterogeneous
Agent Swarms
K.N. Krishnanand, D. Ghose....................................... 165
VI Contents
Self-organization in Evolution for the Solving of Distributed
Terrestrial Transportation Problems
Jean-Charles Cr´eput, Abderrafiaˆa Koukam ........................... 189
Statistical Forecasting of Indian Summer Monsoon Rainfall:
An Enduring Challenge
Shivam Tripathi, Rao S. Govindaraju ................................ 207
Fault Diagnosis of Electronic Circuits Using Cellular
Automata Based Pattern Classifier
Pradipta Maji, P. Pal Chaudhuri.................................... 225
A Survey of Artificial Neural Network-Based Modeling in
Agroecology
Jim´enez Daniel, P´erez-Uribe Andr´es, Satiz´abal H´ector,
Barreto Miguel, Van Damme Patrick, Tomassini Marco................ 247
Software Development Knowledge Management Using
Case-Based Reasoning
Paulo Gomes, Joel Cordeiro, Pedro Gandola, Nuno Seco ............... 271
Learning from Demonstration and Case-Based Planning for
Real-Time Strategy Games
Santiago Ontan˜´on, Kinshuk Mishra, Neha Sugandh, Ashwin Ram ....... 293
A CBR System: The Core of an Ambient Intelligence Health
Care Application
Juan M. Corchado, Javier Bajo, Yanira de Paz ....................... 311
Soft Query-Answering Computing in P2P Systems with
Epistemically Independent Peers
Zoran Majki´c, Bhanu Prasad ....................................... 331
Power Distribution System Fault Diagnosis Using Hybrid
Algorithm of Fuzzy Classification and Artificial Immune
Systems
Le Xu, Mo-Yuen Chow ............................................ 357
Detection of Phishing Attacks: A Machine Learning Approach
Ram Basnet, Srinivas Mukkamala, Andrew H. Sung ................... 373
Author Index................................................... 385
Optimization of Industrial Processes Using Improved and
Modified Differential Evolution
B.V. Babu1 and Rakesh Angira2
1 Chemical Engineering Department & Educational Hardware Division (EHD)
[email protected]
2 Chemical Engineering Department
[email protected]
1,2 Birla Institute of Technology and Science (BITS), Pilani – 333 031 (Rajasthan), India
1 Introduction
Optimization refers to finding one or more feasible solutions, which correspond to ex-
treme values of one or more objectives. The need for finding such optimal solutions in
a problem comes mostly from the extreme purpose of either designing a solution for
minimum possible cost of fabrication, or for maximum possible reliability, or others.
Because of such extreme properties of optimal solutions, optimization methods are of
great importance in practice, particularly in engineering design, scientific experiments
and business decision-making.
In most of the gradient-based traditional optimization methods, in moving along
the gradient, we are guided by the extrapolation of the derivatives of the objective
function with respect to appropriate variables. However, the shape of the response
surface may change, thereby necessitating a change in the direction of search. In other
words, we cannot move on the surface for any considerable length of time. Another
biggest limitation of traditional optimization methods is that these methods are appli-
cable only to find local minima. Hence these traditional optimization techniques do
not ensure the global optimum and also have limited applications.
In the recent years, non-traditional search and optimization methods based on natu-
ral phenomena of evolution, such as Simulated Annealing (SA), Genetic algorithms
(GA), Differential evolution (DE), Self Organizing Migrating Algorithms (SOMA),
Particle Swarm Optimization (PSO), Tabu Search (TS), Scatter Search (SS), and Ant
Colony Optimization (ACO) have been developed to overcome these problems of tra-
ditional optimization methods (Onwubolu and Babu 2004; Babu 2004; Corne et al.
1999). These algorithms are stochastic in nature, with probabilistic transition rule.
These are comparatively new and are gaining popularity due to certain properties
which the deterministic algorithms do not have. These are found to have better global
perspective than the traditional methods.
Previous studies (Storn 1995; Wang and Chiou 1997; Babu & Sastry 1999; Babu
and Angira 2003; Angira and Babu 2005a; Angira 2005) have shown that DE is an ef-
ficient, effective and robust evolutionary optimization method. Still DE takes large
computational time for optimizing the computationally expensive objective functions.
And therefore, an attempt to speed up DE is considered necessary. Modified Differen-
tial Evolution (MDE), an evolutionary optimization technique, is evaluated by apply-
ing to four non-linear chemical processes (Angira and Babu 2005a, 2006a, 2006b,
B. Prasad (Ed.): Soft Computing Applications in Industry, STUDFUZZ 226, pp. 1–22, 2008.
springerlink.com © Springer-Verlag Berlin Heidelberg 2008
2 B.V. Babu and R. Angira
2006c; Babu and Angira 2005, 2006). DE has also been extended to solve multi ob-
jective optimization problems. Three such extended strategies namely MODE (Babu
et al. 2005a, 2005b, 2005c), NSDE (Angira and Babu 2005b) and MNSDE (Angira
and Babu 2006a) are discussed in brief.
2 DE in Brief
Differential Evolution (DE), a recent optimization technique, is an exceptionally sim-
ple evolution strategy, which is significantly faster & robust at numerical optimization
and is more likely to find a function’s true global optimum (Price and Storn 1997).
Simple GA (Goldberg1989; Deb 1996) uses a binary coding for representing problem
parameters whereas DE uses real coding of floating point numbers. Among the DE’s
advantages are its simple structure, ease of use, speed and robustness. The details of
DE algorithm and pseudo code are available in literature (Price and Storn 1997; Babu
and Sastry 1999; Onwubolu and Babu 2004; Angira 2005; Angira and Babu 2006b;
Price and Storn, 2007 etc.).
Original DE dealt with a single strategy (Price and Storn 1997). Later on ten dif-
ferent strategies have been suggested by Price and Storn (2007). A set of control pa-
rameters that works out to be the best for a given problem may not work well when
applied for a different problem. The best value of control parameters to be adopted for
each problem is to be determined separately by trial & error. Similarly, the strategy
that works out to be the best for a given problem may not be effective when applied to
some other problem.
DE has been successfully applied in various fields. Some of the successful applica-
tions of DE include: digital filter design (Storn 1995), Batch fermentation process
(Wang and Chen 1999), Estimation of heat transfer parameters in trickle bed reactor
(Babu and Sastry 1999), Dynamic Optimization of a Continuous Polymer Reactor using
a Modified Differential Evolution (Lee et al. 1999), Optimal design of heat exchangers
(Babu and Munawar 2000), Synthesis and optimization of heat integrated distillation
system (Babu and Singh 2000), Optimization of an alkylation reaction (Babu and Gau-
rav 2000), optimization of Low Pressure Chemical Vapor Deposition Reactors Using
Hybrid Differential Evolution (Lu and Wang 2001), optimization of non-linear func-
tions (Babu and Angira 2001a), Optimization of thermal cracker operation (Babu and
Angira 2001b), Global optimization of MINLP problems (Babu and Angira 2002), Op-
timization of water pumping systems (Babu and Angira 2003), Supply chain Planning
(Pinto 2002), Optimization of Non-Linear Chemical Processes (Babu and Angira 2006),
Process Synthesis and design (Angira and Babu 2006a), Optimal Design of Complex
and Non-Linear Chemical Processes (Angira and Babu 2006b) etc.
3 Improvements on DE
When using any population based search algorithm in general and DE in particular to
optimize a function, an acceptable trade–off between convergence (with reference to
locating optimum) and robustness (with reference to not missing the global optima)
must generally be determined. Convergence implies a fast convergence although it
may be to a local optimum. On the other hand, robustness guarantees a high probabil-
ity of obtaining the global optimum. A few attempts have already been made to
Optimization of Industrial Processes Using Improved and Modified Differential Evolution 3
achieve this trade-off (Chiou and Wang 1999; Babu and Angira 2003; Tasoulis et al
2004; Bergeya and Ragsdaleb 2005).
Chiou and Wang (1999) embedded accelerated phase and migration phase into the
original algorithm of DE. These two phases are used to improve the convergence
speed without decreasing the diversity among individuals. Also, several alternate
methods are compared. Babu and Angira (2003) proposed a variation of mutation and
crossover scheme and investigated its effectiveness by applying to liquid extraction
problem. Tasoulis et al. (2004) explored how Differential Evolution can be parallel-
ized in a virtual parallel environment so as to improve both the speed and the per-
formance of the method. Bergeya and Ragsdaleb (2005) proposed a DE with greedy
random strategy for genetic recombination. They found that modified algorithm has
higher convergence velocity than original DE still maintaining the robustness. In this
paper an attempt has been made to increase the convergence speed of DE without
compromising with the robustness. A modified DE is proposed and evaluated in the
present work to achieve this trade-off.
4 Modified Differential Evolution (MDE)
The principle of modified DE (Angira and Babu 2005a, 2005c; 2006a, 2006b, 2006c;
Babu and Angira 2006) is same as DE. The major difference between DE and MDE is
that MDE maintains only one array. The array is updated as and when a better solu-
tion is found. Also, these newly found better solutions can take part in mutation and
crossover operation in the current generation itself as opposed to DE (where another
array is maintained and these better solutions take part in mutation and crossover op-
erations in next generation). Updating the single array continuously enhances the
convergence speed leading to less function evaluations as compared to DE. This
modification enables the algorithm to get a better trade-off between the convergence
rate and the robustness. By choosing the key parameters (NP, CR, and F)
wisely/appropriately, the problem of premature convergence can be avoided to a large
extent. Such an improvement can be advantageous in many real world problems
where the evaluation of a candidate solution is a computationally expensive operation
and consequently finding the global optimum or a good sub-optimal solution with the
original differential evolution algorithm is too time consuming, or even impossible
within the time available. This has been found to be very true in examples such as op-
timization in the field of computational mechanics, computational magnetics, compu-
tational fluid dynamics and unsteady solidification.
5 Application to Industrial Chemical Processes
The optimization of non-linear constraint problems is relevant to chemical engineer-
ing practice (Floudas 1995). Non-linearities are introduced by process equipment de-
sign relations, by equilibrium relations and by combined heat and mass balances.
There are many chemical processes which are highly nonlinear and complex with ref-
erence to optimal operating conditions with many equality and inequality constraints.
In this paper the following processes are considered for applying DE and MDE: (1)
Optimal operation of alkylation unit, and (2) Heat exchanger network design (3) Re-
actor network design (4) Dynamic optimization of a batch reactor.
4 B.V. Babu and R. Angira
5.1 Optimal Operation of Alkylation Unit
Alkylation process is common in the petroleum industry. A simplified process flow
diagram of an alkylation process is shown in Fig. 1. The process model was described
and solved in literature (Sauer et al. 1964) using successive linear programming. The
process model seeks to determine the optimum set of operating conditions for the
process, based on a mathematical model, which allowed maximization of profit.
Bracken and McCormick (1968) formulated the problem as a direct nonlinear pro-
gramming model with mixed nonlinear inequality and equality constraints and a
nonlinear profit function to be maximized. They used Sequential Unconstrained
Minimization Technique (SUMT) for solving the same. Later, Dembo (1976) trans-
formed the NLP problem with ten variables which Bracken and McCormick (1968)
derived, into a problem with seven variables. All equality constraints are eliminated
and the problem has been formulated as a signomial optimization problem. This prob-
lem involves seven variables subject to twelve nonlinear and two linear inequality
constraints. Edger and Himmelblau (1989) used sequential quadratic programming to
solve the problem as formulated by Bracken and McCormick (1968). Maranas and
Floudas (1997) used generalized geometric programming to solve the seven variables
problem as formulated by Dembo (1976). Adjiman et al. (1998) used αBB algorithm
(for general twice –differentiable constraint NLPs) for solving this problem.
As shown in Fig. 1, an olefin feed (100% butane), a pure isobutane recycle and a
100% isobutane make-up stream are introduced in a reactor together with an acid cata-
lyst. The reactor product stream is then passed through a fractionator where the isobu-
tane and the alkylate product are separated. The spent acid is also removed from the re-
actor. The variables are defined as shown in Table-1 along with the upper and lower
bounds on each variable. The bounds represent economic, physical and performance
constraints. In the present study, the problem formulation is same as that of (Maranas
and Floudas 1997; Adjiman et al. 1998). The problem is briefly discussed below.

Isobutane

Make up
FRACTIONATOR

REACTOR

Olefin feed

Alkylate Product

Fresh Acid
Spent Acid

Fig. 1. Simplified Alkylation Process Flow sheet