Paper Title, Authors, Abstract (Issue 1, Volume 4, 2010)

Pages

New Techniques of Products Analysis
Marius Buzera, Marius-Constantin Popescu, Nikos E. Mastorakis, Jean-Octavian Popescu

Abstract: Researches throughout the past few years, having as a goal the automatic classification of products, via calculus systems implementation, as well as machine vision techniques, and artificial intelligence field methods, have lead to very promising results. Together with the colour, shape is one of the most important parameters of vegetal products. Thus, it helps one learn further information on the integrity of products, information which can be used in their classification, while taking the shape into consideration. Using them allowed for the assessment of some parameters such as shape, colour and the integrity degree of the products analyzed, having much more superior results than the classical classification installations. Still, due to these techniques particularities, the classification process implies going through some more phases. Both the experimental methodology for classifying vegetal products and some original algorithms are presented in this paper. To classify the shape it has been developed back-propagation feed-forward artificial neural network, and for colour a fuzzy algorithm. In order to test these techniques, an experimental device was created to allow a video inspection of products, some of the conclusions being presented in this material.
 

Stabilization of Non-necessarily Inversely Stable First-order Adaptive Systems Under Saturated Input
M. de la Sen, O. Barambones

Abstract: This paper is concerned with an indirect adaptive stabilization scheme for first-order continuous-time systems under saturated input which is described by a sigmoidal function. The control singularities are avoided through a modification estimation scheme for the estimated plant parameter vector so that its associated Sylvester matrix is guaranteed to be non-singular and then the estimated plant model is controllable. This strategy implies at the same time the controllability through time of the modified estimation scheme. The estimation modification mechanism involves the use of a hysteresis switching function. An alternative hybrid scheme, whose estimated parameters are updated at sampling instants is also given to solve a similar adaptive stabilization problem. Such a scheme also uses hysteresis switching for modification of the parameter estimates so as to ensure the controllability of the estimated plant model.
 

    Paper Title, Authors, Abstract (Issue 2, Volume 4, 2010)

Pages

A New Approach to Non-fragile H-infinity Fuzzy Filter of Uncertain Markovian Jump Nonlinear Systems
Wudhichai Assawinchaichote

Abstract: This paper considers the problem of designing a nonfragile H1 fuzzy filter for uncertain Markovian jump nonlinear systems that the guarantees the L2-gain from an exogenous input to an estimate error output being less than or equal to a prescribed value. Sufficient conditions for the existence of the H1 fuzzy filter are given in terms of a set of LMIs. In this paper, the premise variables of the H1 fuzzy filter are allowed to be different from the premise variables of the TS fuzzy model of the plant such that the results are shown into two cases which are the premise variable of the fuzzy model be measurable and the premise variable assumed to be unmeasurable.
 

Methanol Production from Biogas
Anita Kovac Kralj, Davorin Kralj

Abstract: Methanol is produced from synthesis gas, which is produced from natural gas. Natural gas can be replaced by biogas for the production of synthesis gas. We compare the production of methanol from varieties of raw materials - natural gas and biogas. The basic starting point for comparison is the same mass inlet flow rate of both raw materials under the same operating conditions. Methanol production using natural gas and biogas as the raw material was simulated using an Aspen Plus simulator with real chemical thermodynamic, and 16 146 kg/h crude methanol from natural gas and 14 615 kg/h from biogas could be produced. Methanol production from biogas could also increase by 9.7 % with processed operational and parametric modification using nonlinear programming (NLP). The most important is the conversion of methane in the reformer. Optimal methane conversion could take place by operating with the use of optimal parametric data in a reformer unit (temperature=840 oC and pressure=8 bar). The optimal production of methanol from biogas was 16 040 kg/h under optimal parameters.
 

Overview of the 2D and 3D Finite Element Studies versus Experimental Results of a Solid Propellant Engine Performances under Cycling Loading Effect
Adrian Arghiropol, Constantin Rotaru

Abstract: The article will summarize few of the achievements after the experimental and computational research on both 2 D axis symmetric and 3 D axis symmetric Finite Element modeling of the flow inside a solid propellant rocket engine with a specific axial distribution of the propellant’s material temperature generated by the long run flight vibrations. The solid propellant was assumed to be a vascoelastic material under cycling loading. The 2D and 3D modeling results of the rocket engine’s internal flow parameters and performances will be evaluated and compared with few of the performed experimental results.
 

A Quasi Regression Model for Polytomous Data and Its Application for Measuring Service Quality
Alexander Andronov, Nadezda Kolmakova, Irina Yatskiv

Abstract: A quasi nonlinear regression model with polytomous response is considered. Unknown parameters are estimated using maximum likelihood method. Corresponding information matrix is presented. Gotten results are used for an evaluation of transport service quality in Riga Coach Terminal, different examples are considered.
 

    Paper Title, Authors, Abstract (Issue 3, Volume 4, 2010)

Pages

A Comparison of RBF Networks and Random Forest in Forecasting Ozone Day
Hyontai Sug

Abstract: It is known that random forest has good performance for data sets containing some irrelevant features, and it is also known that the performance of random forest is very good at ozone day prediction data set that is supposed to have some irrelevant features. On the other hand, it is known that when data sets do not contain irrelevant features, RBF networks are good at prediction tasks. Moreover, in general, we do not have exact knowledge about irrelevant features, because data space is usually far greater than available data for training. So we want to test that the two facts are true or not for the ozone data set. Experiments were done with random forests and RBF networks using k-means clustering, and showed that RBF networks are slightly better than random forest for the ozone day prediction.
 

GA-Based PIDA Control Design Optimization with an Application to AC Motor Speed Control
Sunisa Sornmuang, Sarawut Sujitjorn

Abstract: PIDA controller has been proposed since 1996 as an extension to the conventional PID controller. The additional term “A” stands for acceleration. With this new term, a closed-loop system can respond faster with less overshoot. Originally, the design utilizes the dominant pole concept proceeded in the s-plane. As shown by simulation, this design approach is not suitable for high-order plants having delays and complex oscillatory modes. The article proposes an algebraic design approach which also utilizes the genetic algorithm (GA) to achieve design optimality. Comparison studies among the previous method, the gradient-search based method and the proposed approach are elaborated. Such studies were conducted against some benchmark plants defined by Astrom and Hagglund. As a result, the GA method with heuristically defined solution boundaries provides superior results. The proposed approach has been successfully applied to the speed control of an AC motor.
 

Hybrid Bacterial Foraging and Tabu Search Optimization (BTSO) Algorithms for Lyapunov’s Stability Analysis of Nonlinear Systems
Suphaphorn Panikhom, Nuapett Sarasiri, Sarawut Sujitjorn

Abstract: This article presents brief descriptions of the bacterial foraging optimization (BFO), the tabu search (TS) and the hybrid algorithms thereof namely bacterial foraging-tabu search optimization (BTSO) algorithms. The proposed hybrid BTSO algorithms perform search rapidly, and render a high-quality solution according to the operation of the adaptive tabu search (ATS). The BTSO algorithm is applied to stability analysis of linear and nonlinear systems based on the Lyapunov’s methods. The stability analysis results are compared with the threshold accepting (TA) method. The article also covers the reviews of the TA and the Lyapunov’s methods, respectively.
 

Stability Conditions for a Retarded Quasipolynomial and their Applications
Libor Pekar, Roman Prokop, Radek Matusu

Abstract: Non-delay real parameter stability and stabilization for a quasipolynomial of a retarded structure is studied in this contribution. In the sense of this paper, quasipolynomials are considered to be over real coefficients and in only one variable. Unlike some other methods and analyses, a non-delay real parameter is being to set in a quasipolynomial with two independent delay terms. Retarded quasipolynomial stability is given by the requirement that all its roots (of an infinite spectrum) are located in the open left-half complex plane. The proposed stabilization methodology is based on the argument principle, i.e. on the Mikhaylov stability criterion. This problem has many applications especially in the control theory since such a quasipolynomial can characterize the dynamics of a closed-loop system with delays. In the presented paper, we introduce two problems connected with stabilization and control of time delay systems. The first one deals with a coprime factorization for algebraic controller design, the second one propose stabilization of an anisochronic model of a high order system. Stability features and application problems are accompanied with simulation examples in the Matlab-Simulink environment.
 

    Paper Title, Authors, Abstract (Issue 4, Volume 4, 2010)

Pages

A Novel Simulation Approach for Analyzing Reactive Molding Process
Robert Rajca, Lukasz Matysiak, Michal Banas, Robert Sekula

Abstract: Reactive molding process of thermosetting materials is an area where advanced computer simulations can provide useful information to detect molding problems prior to the mold making. Examples of such problems are premature gelation, undesired weldline locations and air traps. This paper introduces newly developed simulation approach for analyzing reactive molding processes: filling and curing. The presented methodology differs from the currently used one because does not require deep end-user’s CFD (Computational Fluid Dynamics) knowledge. It was possible thanks to automation of a number of typical in numerical approach stages including: CAD geometry discretization, solving and postprocessing. The presented method starts with CAD geometry preparation in accordance with the set of specific rules. In the next step the geometry is uploaded via the Website. The Web application analyzes the geometry and detects its structure in automated way. These initial information allow creating an individual Website where, in consequence, engineer (end-user responsible for the final process and product design) is able to enter process parameters (e.g. material properties, initial temperatures, velocities, etc.) and start calculations. The discretization (or meshing) and solving stages are performed in fully automated way. Finally the Web application creates the report with useful information helping to understand the phenomena occurring inside the mold. This report is also available via developed Website. Based on these information the engineer makes a decision to accept the design and process parameters or to restart the simulation for further optimization. The presented approach helps to save time required for designing proper product and its mold. It influences directly on one of the most important global market factor means time-to-market and in the same increase product quality.
 

Numerical Analysis of a DFB Fiber Laser Sensor
Sorin Miclos, Dan Savastru, Ion Lancranjan

Abstract: This paper is pointing to numerical simulation of vari-ous aspects of distributed feedback fiber laser sensors and their ap-plications mainly in the field of the aeronautical applications. The developed numerical analysis has the aim of a better understanding of DFB-FL itself and of its interaction with environment in order to be operated as a sensor. Numerical analysis concentrates both on the FEM and phenomenological methods.
 

Theoretical Analysis of the Output Noise of a DFB Fiber Laser Sensor
Dan Savastru, Sorin Miclos, Ion Lancranjan

Abstract: The results of a theoretical analysis of the output noise generated by the DFB fiber laser sensor with applications mainly in the field of the aeronautical applications are presented. The main purpose of this analysis is to evaluate the magnitude of the DFB fiber laser sensor output noise. This evaluation is necessary for a proper design of the sensor, especially regarding sensitivity and dy-namic range. It is demonstrated that the main source of DFB fiber laser sensor output noise is constituted by the Amplified Spontaneous Emission (ASE) of the fiber amplifier. An extended range of linear response was achieved optimizing the sensor parameters. ASE noise level was brought to an acceptable level. Judging the time response, the designed sensor acts like a high fidelity recorder.
 

Cellular Automata Simulation Modeling of HIV Infection in Lymph Node and Peripheral Blood Compartments
Sompop Moonchai, Yongwimon Lenbury, Wannapong Triampo

Abstract: Acquired immune deficiency syndrome (AIDS) has been widely considered as the most devastating epidemic. To discover effective therapy for HIV infection, the dynamics of the virus-immune system in the human body have been the subject of intense studies. Since the development of the disease typically exhibits a three phase evolution, that is, an acute phase (measured in days), a chronic phase (measured in weeks) and AIDS (measured in years), the use of ordinary or partial differential equations are inadequate in our attempt to describe the three different time scales in order to simulate the entire course of the HIV infection. Cellular automata simulation approach has become well known as a useful technique to investigate complex biomedical systems in situations where traditional methodologies are difficult or too costly to employ. So far, relatively simple cellular automata models have been proposed to simulate the dynamics of HIV infection in human. Most cellular automata models only considered viral proliferation in the lymph node. However, most clinical indications of AIDS progression are based on blood data, because these data are most easily obtained. Since viral population circulates between lymph node and plasma, viral load in the two compartments are important for the description of HIV infection dynamics. We present here cellular automata simulations of a two-compartment model of HIV proliferation with delay.
 

Investigation of Spatial Pattern Formation Involving CD4+ T Cells in HIV/AIDS Dynamics by a Stochastic Cellular Automata Model
Monamorn Precharattana, Wannapong Triampo, Charin Modchang, Darapond Triampo, Yongwimon Lenbury

Abstract: In recent years, discrete models have emerged to play an important role in the study of immune response especially in the problem involving human immunodeficiency virus (HIV) infection, leading to AIDS. As infection of target immune cells by HIV mainly takes place in the lymphoid tissue, cellular automata (CA) models thus represent a significant step toward understanding how the infected population is dispersed. Motivated by these considerations, we introduce a stochastic CA model for HIV dynamics and explore the spatiotemporal pattern of infection. The model is successful in reproducing typical evolution of HIV which is observed in the dynamics of CD4+T cells and infected CD+T cells in infected patients. The geographical result on cell distributions illustrates how infected cells can be dispersed by spatial communities. We have found the pattern formation is based on the relationship among cell states, the set of local transition rules, the conditions and the parameters in the systems. The main finding is that the emergence of dead cells barriers greatly controls the pattern formation in our system, by limiting infections and the manner in which the infection dynamics is brought to the last phase after the barrier is destroyed.
 

A Lightweight Method to Parallel K-Means Clustering
Kittisak Kerdprasop, Nittaya Kerdprasop

Abstract: Traditional k-means clustering iteratively performs two major steps: data assignment and calculating the relocation of mean points. The data assignment step sends each data point to a cluster with closest mean, or centroid. Normally, the measure of closeness is the Euclidean distance. On clustering large datasets, the k-means method spends most of its execution time on computing distances between all data points and existing centroids. It is obvious that distance computation of one data point is irrelevant to others. Therefore, data parallelism can be achieved in this case and it is the main focus of this paper. We propose the parallel method as well as its approximation scheme to the k-means clustering. The parallelism is implemented through the message passing model using a concurrent functional language, Erlang. The experimental results show the speedup in computation of parallel k-means. The clustering results of an approximated parallel method are impressive when taking into account its fast running time.
 

Modeling and Analysis of Information Systems Outsourcing based on Agent Systems
Seigo Matsuno, Takao Ito, Masayoshi Hasama, Tatsuo Asai

Abstract: Information systems (IS) outsourcing can be classified into two typical but different patterns: conventional outsourcing and quasi-outsourcing. However, the diversity of the IS outsourcing decision whether firms are increasing, decreasing, or keeping their current outsourcing level is widely seen recently. Therefore, development of mathematical models depending on situations is required to describe and analyze the collaborations/ relationships among firms in a general way. This paper deals with the analysis of profits/prices changes in formalizing of collaboration among agents and understanding of the IS outsourcing phenomena by applying a mathematical model based on agent systems. Up to now, two influential perspectives of outsourcing, that is, the TCE and the RBV have been both making a valuable contribution to understanding and explaining the complexities of outsourcing. However, we revealed that the outcomes of outsourcing can fluctuate inherently according to the degree of the collaboration between firms by our simulation studies. By assuming a firm agent produces goods by using support of another outside agent with several cost of labor usage, then the wealth of a firm agent bears some chaotic fluctuations depending on the rate of collaboration among agents. This finding is applicable to the cases where firms will procure services related to the IS activities from outside vendors in real society. Researchers and practitioners should keep in mind that it is a crucial issue of profits/prices changes according to the degree of collaboration between firms in their IS outsourcing decisions.
 

Ensemble of Duo Output Neural Networks for Binary Classification
Pawalai Kraipeerapun, Somkid Amornsamankul

Abstract: This paper presents an ensemble of duo output neural networks (DONN) using bagging technique to solve binary classification problems. DONN is a neural network that is trained to predict a pair of complementary outputs which are the truth and falsity values. Each component in an ensemble contains two DONNs in which the first network is trained to predict the truth and falsity outputs whereas the second network is trained to predict the falsity and truth outputs which are set in reverse order of the first one. In this paper, we propose classification techniques based on outputs obtained from DONNs. Also, the ensemble selection technique is proposed. This technique is created based on uncertainty and diversity values. All proposed techniques have been tested with three benchmarks UCI data sets, which are ionosphere, pima, and liver. It is found that the proposed ensemble techniques provide better results than those obtained from an ensemble of back propagation neural networks, an ensemble of complementary neural networks, a single pair of duo output neural networks, a single pair of complementary neural networks, and a back propagation neural network.
 

Three-Dimensional Simulation of Femur Bone and Implant in Femoral Canal using Finite Element Method
Somkid Amornsamankul, Kamonchat Kaorapapong, Benchawan Wiwatanapataphee

Abstract: In this paper, a mathematical model is developed to simulate three-dimensional femur bone and femur bone with implant in the femoral canal, taking into account stress distribution and total displacement during horizontal walking. The equilibrium equations are used in the model. Realistic domain are created by using CT scan data. Different cases of static loads and different boundary conditions are used in the simulation. The finite element method is utilized to determine total displacement and Von Mises Stress. The influences of human weight during horizontal walking are investigated. This model will give the useful for surgeon in femur surgeries. The results show that higher weight provides higher total displacement. And it is found that the Von Mises stress affects the lateral femur.