• Journal of Institute of Control, Robotics and Systems
• /
• v.17 no.8
• /
• pp.731-738
• /
• 2011

In this paper, we introduce an advanced architecture of K-Means clustering-based polynomial Radial Basis Function Neural Networks (p-RBFNNs) designed with the aid of SSOA (Space Search Optimization Algorithm) and develop a comprehensive design methodology supporting their construction. In order to design the optimized p-RBFNNs, a center value of each receptive field is determined by running the K-Means clustering algorithm and then the center value and the width of the corresponding receptive field are optimized through SSOA. The connections (weights) of the proposed p-RBFNNs are of functional character and are realized by considering three types of polynomials. In addition, a WLSE (Weighted Least Square Estimation) is used to estimate the coefficients of polynomials (serving as functional connections of the network) of each node from output node. Therefore, a local learning capability and an interpretability of the proposed model are improved. The proposed model is illustrated with the use of nonlinear function, NOx called Machine Learning dataset. A comparative analysis reveals that the proposed model exhibits higher accuracy and superb predictive capability in comparison to some previous models available in the literature.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.7 no.2
• /
• pp.96-105
• /
• 1997

In this paper, A new design method of fuzzy modeling is presented for the model identification of nonlinear complex systems. The proposed FPNN(Fuzzy Polynomial Neural Network) modeling implements system structure and parameter identification using GMDH(Group Method of Data Handling) method and linguistic fuzzy implication rules from input and output data of processes. In order to identify premise structure and parameter of fuzzy implication rules, GMDH method and regression polynomial fuzzy reasoning method are used and the least square method is utilized for the identification of optimum consequence parameters. Time series data for gas furnace and those for wastewater treatment process are used for the purpose of evaluating the performance of the proposed FPNN modeling. The results show that the proposed method can produce the fuzzy model with higher accuracy than other works achieved previously.

• Atmosphere
• /
• v.18 no.1
• /
• pp.1-14
• /
• 2008

The cloud amount, one of the basic parameter in atmospheric observation, have been observed by naked eyes of observers, which is affected by the subjective view. In order to ensure reliable and objective observation, a new algorithm to retrieve cloud amount was constructed using true color images composed of red, green and blue (RGB). The true color image is obtained by the Skyview, an all-sky imager taking pictures of sky, at the Science Building of Yonsei University, Seoul for a year in 2006. The principle of distinguishing clear sky from cloudy sky lies in the fact that the spectral characteristics of light scattering is different for air molecules and cloud. The result of Skyview's algorithm showed about 77% agreement between the observed cloud amount and the calculated, for the error range, the difference between calculated and observed cloudiness, within ${\pm}2$. Seasonally, the best accuracy of about 83% was obtained within ${\pm}2$ range in summer when the cloud amounts are higher, thus better signal-to-noise ratio. Furthermore, as the sky turbidity increased, the error also increased because of increased scattering which can explain the large error in spring. The algorithm still need to be improved in classifying sky condition more systematically with other complimentary instruments to discriminate thin cloud from haze to reduce errors in detecting clouds.

• Steel and Composite Structures
• /
• v.27 no.3
• /
• pp.311-325
• /
• 2018

This article presents the bending analysis of FGM rectangular plates resting on non-uniform elastic foundations in thermal environment. Theoretical formulations are based on a recently developed refined shear deformation theory. The displacement field of the present theory is chosen based on nonlinear variations in the in-plane displacements through the thickness of the plate. The present theory satisfies the free transverse shear stress conditions on the top and bottom surfaces of the plate without using shear correction factor. Unlike the conventional trigonometric shear deformation theory, the present refined shear deformation theory contains only four unknowns as against five in case of other shear deformation theories. The material properties of the functionally graded plates are assumed to vary continuously through the thickness, according to a simple power law distribution of the volume fraction of the constituents. The elastic foundation is modeled as non-uniform foundation. The results of the shear deformation theories are compared together. Numerical examples cover the effects of the gradient index, plate aspect ratio, side-to-thickness ratio and elastic foundation parameters on the thermo-mechanical behavior of functionally graded plates. Numerical results show that the present theory can archive accuracy comparable to the existing higher order shear deformation theories that contain more number of unknowns.

• Structural Engineering and Mechanics
• /
• v.65 no.1
• /
• pp.33-41
• /
• 2018

In this paper, using modified couple stress theory in place of classical continuum theory, and using shell model in place of beam model, vibrational behavior of nanotubes is investigated via the finite element method. Accordingly classical continuum theory is unable to correctly compute stiffness and account for size effects in micro/nanostructures, higher order continuum theories such as modified couple stress theory have taken on great appeal. In the present work the mass-stiffness matrix for cylindrical shell element is developed, and by means of size-dependent finite element formulation is extended to more precisely account for nanotube vibration. In addition to modified couple stress cylindrical shell element, the classical cylindrical shell element can also be defined by setting length scale parameter to zero in the equations. The boundary condition were assumed simply supported at both ends and it is shown that the natural frequency of nano-scale shell using the modified coupled stress theory is larger than that using the classical shell theory and the results of Ansys. The results have indicated using the modified couple stress cylindrical shell element, the rigidity of the nano-shell is greater than that in the classical continuum theory, which results in increase in natural frequencies. Besides, in addition to reducing the number of elements required, the use of this type of element also increases convergence speed and accuracy.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.12 no.1
• /
• pp.454-475
• /
• 2018

This paper presents a novel Android malware classification model planned to classify and categorize Android malicious code at Drebin dataset. The amount of malicious mobile application targeting Android based smartphones has increased rapidly. In this paper, Restricted Boltzmann Machine and Deep Belief Network are used to classify malware into families of Android application. A texture-fingerprint based approach is proposed to extract or detect the feature of malware content. A malware has a unique "image texture" in feature spatial relations. The method uses information on texture image extracted from malicious or benign code, which are mapped to uncompressed gray-scale according to the texture image-based approach. By studying and extracting the implicit features of the API call from a large number of training samples, we get the original dynamic activity features sets. In order to improve the accuracy of classification algorithm on the features selection, on the basis of which, it combines the implicit features of the texture image and API call in malicious code, to train Restricted Boltzmann Machine and Back Propagation. In an evaluation with different malware and benign samples, the experimental results suggest that the usability of this method---using Deep Belief Network to classify Android malware by their texture images and API calls, it detects more than 94% of the malware with few false alarms. Which is higher than shallow machine learning algorithm clearly.

• Transactions of Materials Processing
• /
• v.26 no.1
• /
• pp.11-17
• /
• 2017

In this study, process analysis of elbow-shaped tubes using a mandrel has been performed. To reach the final shape within the dimensional tolerance, the process analysis has been performed at various processing parameters such as tube dimensions, the curved cutting surface and the radius of curvature. The area outside the boundary of the target shape was expressed as a quantitative index to analyze the formability. The validation experiments have also been performed in order to increase the reliability of the process analysis. For the processing of elbow-shaped tubes, it is preferable to make the angle of the portion where the punch touches the tube smaller than the opposite angle. And the convex cutting surface is advantageous due to the increased contacts between the punch and the tube ends during the bending process. Elbow tube having larger radius of curvature shows higher dimensional accuracy due to the relatively uniform strain distribution.

• Journal of Mechanical Science and Technology
• /
• v.18 no.1
• /
• pp.45-54
• /
• 2004

In order to meet the requirement for higher thickness accuracy in cold rolling processes, it is strongly desired to have high performance in control units. To meet this requirement, we have considered an output regulating control system with a roll-eccentricity estimator for each rolling stand of tandem cold mills. Considering entry thickness variation as well as roll eccentricity as the major disturbances, a synthesis of multivariable control systems is presented based on H$\sub$$\infty$/ control theory, which can reflect the knowledge of input direction and spectrum of disturbance signals on the design. Then, to reject roll eccentricity effectively, a weight function having some poles on the imaginary axis is introduced. This leads to a non-standard H_ control problem, and the design procedures for solving this problem are analytically presented. The effectiveness of the proposed control method is evaluated through computer simulations and compared to that of the conventional LQ control and feedforward control methods for roll eccentricity.

• Journal of Advanced Marine Engineering and Technology
• /
• v.21 no.4
• /
• pp.430-436
• /
• 1997

Time-dependent Flow characteristics of a two-dimensional lid-driven square cavity flow of six high Reynolds numbers, $10^4 2{\times}lO^4., 3{\times}l0^4, 5{\times}lO^4, 7.5{\times}lO^4$ and $10^5$ were investigated. A convection conservative difference scheme based upon SOLA to maintain the nearly 2nd-order spatial accuracy was adopted on irregular grid formation. Irregular grid number is $80{\times}80$ and its minimum size is about 1/400 of the cavity height(H) and its maximum is about 1/53 H. The result shows that at Re= $10^4$, periodic migration of small eddies appearing in corner separation region and its temporal sinusoidal fluctuation are represented. And another critical Reynolds number which shows the commencement of flow randomness emerging from the periodic fluctuation is assumed to be around Re= $1.5{\times}10$. At five higher Reynolds numbers ($2{\times}lO^4., 3{\times}l0^4, 5{\times}lO^4, 7.5{\times}lO^4$ and $10^5$), an organizing structure of four consecutive vortices similar to a Moffat vortex at two lower corners is revealed from time-mean flow patterns.

• Proceedings of the Materials Research Society of Korea Conference
• /
• 2010.05a
• /
• pp.19.1-19.1
• /
• 2010

The process of manufacturing printed electronics using printing technology is attracting attention because its process cost is lower than that of the conventional semiconductor process. This technology, which offers both a lower cost and higher productivity, can be applied in the production of organic TFT (thin film transistor), solar cell, RFID(radio frequency identification) tag, printed battery, E-paper, touch screen panel, black matrix for LCD(liquid crystal display), flexible display, and so forth. In general, in order to implement printed electronics, narrow width and gap printing, registration of multi-layer printing by several printing units, and printing accuracy of under $20\;{\mu}m$ are all required. These electronic products require high precision to the degree of tens of microns - in a large area with flexible material, and mass productivity at low cost. As such, the roll-to-roll printing process is attracting attention as a mass production system for these printed electronic devices. For the commercialization of this process, two basic electronic ink technologies, such as conductive ink and polymers, and printing equipment have to be developed. Therefore, this paper addressed basis design and test to develop fine patterning equipment employing the roll-to-roll printing equipment and electronic ink.