• 한국지능시스템학회:학술대회논문집
• /
• 한국퍼지및지능시스템학회 2004년도 추계학술대회 학술발표 논문집 제14권 제2호
• /
• pp.313-316
• /
• 2004

Model selection is the process that sets up the regularization parameter in the support vector machine or regularization network by using the external methods such as general cross validation or L-curve criterion. This paper suggests that the regularization parameter can be obtained simultaneously within the learning process of neural networks without resort to separate selection methods. In this paper, extended kernel method is introduced. The relationship between regularization parameter and the bias term in the extended kernel is established. Experimental results show the effectiveness of the new model selection method.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2008년도 하계종합학술대회
• /
• pp.773-774
• /
• 2008

As an Embedded system get more complicated, the importance of the software test increases. Especially it is mandatory to test the embeded software for a charge degrade of the development company which is happened due to the error, a danger element clearance to the customer, a performance enhance of the system. In this paper, we propose efficient test method which discovers the bug an embedded system using suitable test case selection and automation test.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1989년도 한국자동제어학술회의논문집; Seoul, Korea; 27-28 Oct. 1989
• /
• pp.705-710
• /
• 1989

A practical method of identifying the inertial parameters, viscous friction and Coulomb friction of a robot is presented. The parameters in the dynamic equations of a robot are obtained from the measurements of the command voltage and the joint position of the robot. First, a dynamic model of the integrated motor and manipulator is derived. An off line parameter identification procedure is developed and applied to the University of Minnesota Direct Drive Robot. To evaluate the accuracy of the parameters the dynamic tracking of robot was tested. The trajectory errors were significantly reduced when the identified dynamic parameters were used.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 2003년도 춘계학술대회 논문집
• /
• pp.185-190
• /
• 2003

At the present, a stud bolt welding was achieved by the manual method. The manual method caused many problems of work evasion. In order to work out these problems, an automatic welding system is designed for a stud bolt welding system. The system is composed of articulated type welding robot, stud gun, stud feeder, stud controller and various jig.

• International Journal of Control, Automation, and Systems
• /
• 제1권4호
• /
• pp.511-514
• /
• 2003

The Infrared Camera usually detects only Infrared waves emitted from the light in order to illustrate the temperature distribution. An Infrared diagnosis system can be applied to various fields. But the defect discrimination can be automatic or mechanized in the special shoes total inspection system. This study introduces a method for special shoes nondestructive total inspection. Performance of the proposed method is shown through thermo-Image.

• 한국기계연구소 소보
• /
• 통권19호
• /
• pp.141-147
• /
• 1989

Using monochrome CCD camera with high speed electronic shutter. blade image is captured in the environment with vibration. The image is preprocessed using LoG filter and zero crossing. We used Hough transformation method to detect straight lines and obtained width between lines caused by grinders. This paper proposes automatic blade measuring method to automate the process control of blade grinding.

• International Journal of Control, Automation, and Systems
• /
• 제4권5호
• /
• pp.539-544
• /
• 2006

An $H_2$ filter is derived for time delay systems, where there are time delay terms in the state and in the output. A method to compute the $H_2$ norm of time delay systems is proposed. Based on the $H_2$ norm computation method, an $H_2$ filter design is formulated as a nonlinear optimization problem.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 1998년도 추계학술대회 논문집 학회본부 B
• /
• pp.543-545
• /
• 1998

Robotic soccer is multi-agent system playing soccer game under given rule. This system consists of three mobile robots, vision sensor, action decision module, action execution module and communication module. This paper presents new action decision method using multi-layer neural networks.

• International Journal of Control, Automation, and Systems
• /
• 제6권5호
• /
• pp.677-688
• /
• 2008

In this paper, the general problem of impedance control for a robotic manipulator with a moving flexible base is addressed. Impedance control imposes a relation between force and displacement at the contact point with the environment. The concept of impedance control of flexible base mobile manipulator is rather new and is being considered for first time using singular perturbation and new sliding mode control methods by authors. Initially slow and fast dynamics of robot are decoupled using singular perturbation method. Slow dynamics represents the dynamics of the manipulator with rigid base. Fast dynamics is the equivalent effect of the flexibility in the base. Then, using sliding mode control method, an impedance control law is derived for the slow dynamics. The asymptotic stability of the overall system is guaranteed using a combined control law comprising the impedance control law and a feedback control law for the fast dynamics. As first time, base flexibility was analyzed accurately in this paper for flexible base moving manipulator (FBMM). General dynamic decoupling, whole system stability guarantee and new composed robust control method were proposed. This proposed Sliding Mode Impedance Control Method (SMIC) was simulated for two FBMM models. First model is a simple FBMM composed of a 2 DOFs planar manipulator and a single DOF moving base with flexibility in between. Second FBMM model is a complete advanced 10 DOF FBMM composed of a 4 DOF manipulator and a 6 DOF moving base with flexibility. This controller provides desired position/force control accurately with satisfactory damped vibrations especially at the point of contact. This is the first time that SMIC was addressed for FBMM.

• Smart Structures and Systems
• /
• 제16권2호
• /
• pp.281-294
• /
• 2015

Bolted joint connection is the most commonly used connection element in structures and devices. The loosening due to external dynamic loads cannot be observed and measured easily and may cause catastrophic loss especially in an extreme requirement and/or environment. In this paper, an innovative Real-time Cross-Correlation Method (RCCM) for monitoring of the bolted joint loosening was proposed. We apply time reversal process on stress wave propagation to obtain correlation signal. The correlation signal's peak amplitude represents the cross-correlation between the loosening state and the baseline working state; therefore, it can detect the state of loosening. Since the bolt states are uncorrelated with noise, the peak amplitude will not be affected by noise and disturbance while it increases SNR level and increases the measured signals' reliability. The correlation process is carried out online through physical wave propagation without any other post offline complicated analyses and calculations. We implemented the proposed RCCM on a single bolt/nut joint experimental device to quantitatively detect the loosening states successfully. After that we implemented the proposed method on a real large structure (reaction wall) with multiple bolted joint connections. Loosening indexes were built for both experiments to indicate the loosening states. Finally, we demonstrated the proposed method's great anti-noise and/or disturbance ability. In the instrumentation, we simply mounted Lead Zirconium Titanate (PZT) patches on the device/structure surface without any modifications of the bolted connection. The low-cost PZTs used as actuators and sensors for active sensing are easily extended to a sensing network for large scale bolted joint network monitoring.