Kure National College of Technology

Hiroshima, Japan

Kure National College of Technology

Hiroshima, Japan
SEARCH FILTERS
Time filter
Source Type

Sugitani T.,Hiroshima University | Kubota S.,Hiroshima University | Toya A.,Kure National College of Technology | Xiao X.,Tianjin University | Kikkawa T.,Hiroshima University
IEEE Antennas and Wireless Propagation Letters | Year: 2013

A compact 4 ×: 4 planar ultrawideband (UWB) antenna array with the total size of 44,×:, 52.4 mm2 was developed for radar-based breast cancer detection system. The center frequency and the bandwidth of the antenna were 6 and 12.5 GHz, respectively. The breast phantom materials were developed to fit the characteristics of the measured human breast tissues. A quasi-three- dimensional confocal imaging was performed using the breast phantoms. It was confirmed that the compact 4 ×: 4 antenna array could detect a 5×: 5×: 5-mm3 tumor phantom in an inhomogeneous structure with a glandular phantom and resolve the two separate tumor phantoms, which were located at the depth of 23 mm with the spacing of 10 mm. © 2002-2011 IEEE.


Yoshitake I.,Yamaguchi University | Ogawa A.,Ube Machinery Co. | Kim Y.J.,University of Colorado at Denver | Mimura Y.,Kure National College of Technology
Journal of Structural Engineering (United States) | Year: 2012

This paper presents an experimental study to develop a new composite slab system consisting of a concrete deck with reinforcing bars, steel plates welded with rib connectors, and a carbon fiber-blended cementitious adhesive. The adhesive has unique characteristics such as rust resistance and being waterproof and can be sprayed onto a steel substrate. The foci of the present research are bond performance of the adhesive, which is an interfacial medium between the concrete and steel plates, bond-slip response of the interface, and structure-level investigations. Test parameters include the presence of an adhesive layer, the age of the concrete-steel interface linked with the adhesive, and the configuration of rib connectors. Adequate curing time for the concrete-steel interface with the adhesive is found to take at least 7 days to achieve reasonable composite action. The adhesive layer significantly improves the performance of the interface, including bond slip and load-carrying capacities. The shape of perforation in the rib is not a significant factor on the behavior of the interface. Although the proposed composite concept is promising, progressive bond deterioration is an important consideration, and further research is necessary. © 2012 American Society of Civil Engineers.


Yoshitake I.,Yamaguchi University | Zhang W.,Ningxia University | Mimura Y.,Kure National College of Technology | Saito T.,Materia Inc.
Construction and Building Materials | Year: 2013

The present study investigates the tensile strength and Young's modulus of fly-ash concrete at early age by using a prismatic specimen with an embedded reinforcing bar. Tests were conducted on both normal concrete and fly-ash concrete with a cement-replacement ratio of 20%. Three fly-ash concretes were placed and cured in field environmental conditions in summer, autumn and winter in western Japan. The tension test was performed to quantify the uniaxial tensile strength and tensile Young's modulus of such fly-ash concrete at early age. Based on the test results, the paper presents empirical formulae for both strength and Young's modulus. © 2012 Elsevier Ltd. All rights reserved.


Inoue H.,Kure National College of Technology
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2014

Recently, multiple classifier systems have been used for practical applications to improve classification accuracy. Self-generating neural networks (SGNN) are one of the most suitable base-classifiers for multiple classifier systems because of their simple settings and fast learning ability. However, the computation cost of the multiple classifier system based on SGNN increases in proportion to the numbers of SGNN. In this paper, we propose a novel pruning method for efficient classification and we call this model a self-organizing neural grove (SONG). Experiments have been conducted to compare the SONG with bagging and the SONG with boosting, and support vector machine (SVM). The results show that the SONG can improve its classification accuracy as well as reducing the computation cost. ©Springer International Publishing Switzerland 2014.


Yokose Y.,Kure National College of Technology
Artificial Life and Robotics | Year: 2016

Global warming and environmental destruction are caused in part by the mass consumption of energy by industries that use robotic manipulators. Hence, there is a need to minimize the energy used for manipulator control systems. It is relatively easy to analytically obtain an optimal solution for a linear system. However, a multi-link manipulator is governed by a nonlinear dynamical equation that is difficult to solve as a two-point boundary value problem. Here, the manipulator angles are approximated by Taylor and Fourier series, whose coefficients are sought by a genetic algorithm (GA) to optimize the objective function subject to the boundary conditions. A search method is proposed for planning the trajectory of a manipulator with nonlinear friction and geometrical constraints. © 2016 ISAROB


Yoshikawa Y.,Kure National College of Technology
Proceedings of the Asia and South Pacific Design Automation Conference, ASP-DAC | Year: 2013

A binding method in high-level synthesis for path delay testability is proposed in this paper. For a given scheduled data flow graph, the proposed method synthesizes a path delay testable RTL datapath and its controller. Every path in the datapath is two pattern testable with the controller if the path is activated in the functional operation, i.e., the path is not false path. Our experimental results show that the proposed method can synthesize such RTL circuits with small area overhead compared with that augmented by some DFT techniques such as scan design. © 2013 IEEE.


Inoue H.,Kure National College of Technology
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) | Year: 2013

Recently, multiple classifier systems have been used for practical applications to improve classification accuracy. Self-generating neural networks (SGNN) are one of the most suitable base-classifiers for multiple classifier systems because of their simple settings and fast learning ability. However, the computation cost of the multiple classifier system based on SGNN increases in proportion to the numbers of SGNN. In this paper, we propose a novel pruning method for efficient classification and we call this model a self-organizing neural grove (SONG). Experiments have been conducted to compare the SONG with bagging and the SONG with boosting, the multiple classifier system based on C4.5, and support vector machine (SVM). The results show that the SONG can improve its classification accuracy as well as reducing the computation cost. Additionally, we investigate SONG's incremental learning performance. © Springer-Verlag 2013.


Mori S.,Kure National College of Technology
Precision Engineering | Year: 2015

Bessel beams can be used for high aspect ratio laser drilling and for eliminating the need to precisely position materials along the propagation direction during laser drilling and cutting. However, Bessel beams have side lobes that can damage the materials subjected to these beams. This paper discusses optical suppression of side lobes. A method is proposed to suppress these side lobes, and the method is based on interference between two Bessel beams with different wave vectors. The effectiveness of this method is confirmed both theoretically and experimentally by realizing a superposed Bessel beam; using a He-Ne laser (λ = 633 nm) and an annular binary aperture placed in front of a convex lens, this beam has a 1/e2 radius of 44 μm. © 2014 Elsevier Inc.


Ichinose K.,Kure National College of Technology | Kuroki F.,Kure National College of Technology
2011 China-Japan Joint Microwave Conference Proceedings, CJMW 2011 | Year: 2011

An FPGA-based digital signal processing circuit to detect velocities of targets was considered for NRD guide pulse radar at 60 GHz. Because what to keep in mind is using a previously-designed digital signal processing circuit to detect distances to the targets from the radar, we add the circuit to a function to easily detect the velocities of the targets. In this paper, the digital signal processing circuit was developed to detect the velocities of the multi targets by differentiating the measured distances of those. To confirm the operation of the designed circuit, a pseudo reflection pulse train from the targets was made and inputted into the circuit. As a result, the velocities detection was successfully performed according to the theory. © 2011 CJMW.


Tanaka T.,Kure National College of Technology | Kuroki F.,Kure National College of Technology
European Microwave Week 2013, EuMW 2013 - Conference Proceedings; EuMC 2013: 43rd European Microwave Conference | Year: 2013

In a self-injection locked NRD guide Gunn oscillator using a high-Q ceramic resonator, a sudden change of an oscillation frequency from a locking state to a free-running state was often observed. In this paper, to understand this phenomenon in the reflection type of self-injection locked NRD guide Gunn oscillator, we theoretically and experimentally investigated a condition that such a mode-jump may occur as a parameter of a frequency deviation between the resonant frequency of the ceramic resonator and the oscillation frequency of the free-running oscillator. © 2013 European Microwave Association.

Loading Kure National College of Technology collaborators
Loading Kure National College of Technology collaborators