• Proceedings of the KWS Conference
• /
• 2005.11a
• /
• pp.144-146
• /
• 2005

Dual electromagnetic sensor is used for sensing the weld line. The sensor consists of excitation and two sensing coil wound over the ferro-magnetic core. By using the dual sensor, the effect of noise is minimized. It is based on the generation of eddy currents in the welding plate by passing current through the excitation coil. The sensor can be used to track the butt joints having no gap between them, where a vision based sensor fails to track. Sensor sensitivity depends on the number of coil turns, frequency of excitation, distance of a sensor from the work piece, diameter of core, etc. The whole system consists of a sensor, a signal processing board, a motion controller and a personnel computer (PC). The raw sensor signal is processed using the signal processing board. It consists of amplification, rectification, filtering, averaging, offset adjustment, etc. Based on sensor data, the motion controller adjusts the position of a welding torch.

• Journal of Institute of Control, Robotics and Systems
• /
• v.20 no.10
• /
• pp.1051-1056
• /
• 2014

In this paper, we introduce an intelligent robotic arc welding system which exploits sensors like as LVS (Laser Vision Sensor), Hall effect sensor, voltmeter and so on. The use of industrial robot is saturated because of its own limitation, and one of the major limitations is that industrial robot cannot recognize the environment. Lately, sensor-based environmental awareness research of the industrial robot is performed actively to overcome such limitation, and it can expand application field and improve productivity. We classify the sensor-based intelligent arc welding robot system by the goal and the sensing data. The goals can be categorized into detection of a welding start point, tracking of a welding line and correction of a torch deformation. The Sensing data can be categorized into welding data (i.e. current, voltage and short circuit detection) and displacement data (i.e. distance, position). This paper covers not only the explanation of the each category but also its advantage and limitation.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2012.08a
• /
• pp.268-268
• /
• 2012

Atmospheric pressure microwave induced plasmas are used to excite and ionize chemical species for elemental analysis, for plasma reforming, and for plasma surface treatment. Microwave plasma differs significantly from other plasmas and has several interesting properties. For example, the electron density is higher in microwave plasma than in radio-frequency (RF) or direct current (DC) plasma. Several types of radical species with high density are generated under high electron density, so the reactivity of microwave plasma is expected to be very high [1]. Therefore, useful applications of atmospheric pressure microwave plasmas are expected. The surface characteristics of SUS304 stainless steel are investigated before and after surface modification by microwave plasma under atmospheric pressure conditions. The plasma device was operated by power sources with microwave frequency. We used a device based on a coaxial transmission line resonator (CTLR). The atmospheric pressure plasma jet (APPJ) in the case of microwave frequency (880 MHz) used Ar as plasma gas [2]. Typical microwave Pw was 3-10 W. To determine the optimal processing conditions, the surface treatment experiments were performed using various values of Pw (3-10 W), treatment time (5-120 s), and ratios of mixture gas (hydrogen peroxide). Torch-to-sample distance was fixed at the plasma edge point. Plasma treatment of a stainless steel plate significantly affected the wettability, contact angle (CA), and free energy (mJ/$m^2$) of the SUS304 surface. CA and ${\gamma}$ were analyzed. The optimal surface modification parameters to modify were a power of 10 W, a treatment time of 45 s, and a hydrogen peroxide content of 0.6 wt% [3]. Under these processing conditions, a CA of just $9.8^{\circ}$ was obtained. As CA decreased, wettability increased; i.e. the surface changed from hydrophobic to hydrophilic. From these results, 10 W power and 45 s treatment time are the best values to minimize CA and maximize ${\gamma}$.

• Journal of Welding and Joining
• /
• v.27 no.2
• /
• pp.32-37
• /
• 2009

In shipyard, plate forming is widely used to form the ship hull plate in various shapes. Line heating method by using a flame torch is one of the major shipbuilding processes carried out by skilled workers. Since the forming characteristics depend upon their experiences in manual forming, there are much variations between products and difficulties in communication between engineers and workers. Hence, it needs to develop an automatic forming system which can not only reduce the working time and rework costs but also improve the working environment and hull forming productivity. One of the final goals of plate forming automation is to form a target shape from the initial plate automatically. For automated plate forming, it is required to determine where and how to heat on the plate. To realize this procedure, the inverse problem should be first solved and the effect of curvature shape formed at the heating path should be investigated. In this study, the inverse problem was solved by geometrical approach using the relationship between bending angle and radius of curvature of the curved shape. In addition, experiments of two-dimensional plate forming were performed with the distance-based method considering the curved bending with curvature. The result of the formed shape agreed considerably well with the target shape.

• Transactions of the Korean Society of Mechanical Engineers B
• /
• v.20 no.5
• /
• pp.1671-1678
• /
• 1996

In external chemical vapor deposition processes including VAD and OVD the distribution of flame-synthesized silica particles is determined by heat and mass transfer limitations to particle formation. Combustion gas flow velocities are such that the particle diffusion time scale is longer than that of gas flow convection in the zone of particle formation. The consequence of these effects is that the particles formed tend to remain along straight smooth flow stream lines. Silica particles are formed due to oxidation and hydrolysis. In the hydrolysis, the particles are formed in diffuse bands and particle formation thus requires the diffusion of SiCl$\_$4/ toward CH$\_$4//O$\_$2/ combustion zone to react with H$\_$2/O diffusing away from these same zones on the torch face. The conversion kinetics of hydrolysis is fast compared to diffusion and the rate of conversion is thus diffusion-limited. In the language of combustion, the hydrolysis occurs as a Burke-Schumann process. In selected conditions, reaction zone shape and temperature distributions predicted by the Burke-Schumann analysis are introduced and compared with experimental data available. The calculated centerline temperatures inside the reaction zone agree well with the data, but the calculated values outside the reaction zone are a little higher than the data since the analysis does not consider diffusion in the axial direction and mixing of the combustion products with ambient air. The temperatures along the radial direction agree with the data near the centerline, but gradually diverge from the data as the distance is away from the centerline. This is caused by the convection in the radial direction, which is not considered in the analysis. Spatial distribution of silica particles are affected by convection and diffusion, resulting in a Gaussian form in the radial direction.

• Journal of the Korean Society for Precision Engineering
• /
• v.24 no.10
• /
• pp.64-74
• /
• 2007

Because of the size and environment in the shipbuilding process, the portable type robot is required for the automatic seam tracking. For this reason, the structure of laser sensor should be considered in the initial design step and the coordinate transformation between welding robot and laser sensor, which is joint finder, must be identified exactly and the real time tracking algorithm based on these consideration could be developed. In this research, laser displacement sensor in which its structure is laser beam's vertical projection, is developed to recognize the location of weld joint. In practical applications, however, images of weld joints are often degraded because of the surface specularity or spatter. To overcome the problem, the constrained joint finding algorithm is proposed. In the approach of coordinate conversion rule for the visual feedback control among welding torch, robot body and laser sensor is applied by the same reference point method. In the real time seam tracking algorithms we propose constrained sampling method which uses look ahead distance. The RLS(Recursive Least Square) filter is applied to obtain the smooth tracking path from the sensitive edge data. From the experimental results, we could see the possibility that the developed laser sensor with proposed processing algorithm and real time seam tracking method can be used as a welding under the shipbuilding condition.

• Convergence Security Journal
• /
• v.23 no.4
• /
• pp.61-70
• /
• 2023

Clustering is an unsupervised learning method that involves grouping data based on features such as distance metrics, using data without known labels or ground truth values. This method has the advantage of being applicable to various types of data, including images, text, and audio, without the need for labeling. Traditional clustering techniques involve applying dimensionality reduction methods or extracting specific features to perform clustering. However, with the advancement of deep learning models, research on deep clustering techniques using techniques such as autoencoders and generative adversarial networks, which represent input data as latent vectors, has emerged. In this study, we propose a deep clustering technique based on deep learning. In this approach, we use an autoencoder to transform the input data into latent vectors, and then construct a vector space according to the cluster structure and perform k-means clustering. We conducted experiments using the MNIST and Fashion-MNIST datasets in the PyTorch machine learning library as the experimental environment. The model used is a convolutional neural network-based autoencoder model. The experimental results show an accuracy of 89.42% for MNIST and 56.64% for Fashion-MNIST when k is set to 10.