• The Journal of Korea Robotics Society
• /
• v.18 no.4
• /
• pp.436-443
• /
• 2023

With the advancement of robot intelligence, the concept of man and unmanned teaming (MUM-T) has garnered considerable attention in military research. In this paper, we present a transformer-based architecture for predicting the health status of agents, with the help of multi-head attention mechanism to effectively capture the dynamic interaction between friendly and enemy forces. To this end, we first introduce a framework for generating a dataset of battlefield situations. These situations are simulated on a virtual simulator, allowing for a wide range of scenarios without any restrictions on the number of agents, their missions, or their actions. Then, we define the crucial elements for identifying the battlefield, with a specific emphasis on agents' status. The battlefield data is fed into the transformer architecture, with classification headers on top of the transformer encoding layers to categorize health status of agent. We conduct ablation tests to assess the significance of various factors in determining agents' health status in battlefield scenarios. We conduct 3-Fold corss validation and the experimental results demonstrate that our model achieves a prediction accuracy of over 98%. In addition, the performance of our model are compared with that of other models such as convolutional neural network (CNN) and multi layer perceptron (MLP), and the results establish the superiority of our model.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.11
• /
• pp.81-87
• /
• 2023

Intrusion detection systems that learn metadata of network packets have been proposed recently. However these approaches require time to analyze packets to generate metadata for model learning, and time to pre-process metadata before learning. In addition, models that have learned specific metadata cannot detect intrusion by using original packets flowing into the network as they are. To address the problem, this paper propose a natural language processing-based intrusion detection system that detects intrusions by learning the packet payload as a single sentence without an additional conversion process. To verify the performance of our approach, we utilized the UNSW-NB15 and Transformer models. First, the PCAP files of the dataset were labeled, and then two Transformer (BERT, DistilBERT) models were trained directly in the form of sentences to analyze the detection performance. The experimental results showed that the binary classification accuracy was 99.03% and 99.05%, respectively, which is similar or superior to the detection performance of the techniques proposed in previous studies. Multi-class classification showed better performance with 86.63% and 86.36%, respectively.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.22 no.3
• /
• pp.185-192
• /
• 2017

This paper proposes a single-stage interleaved soft-switching electrolytic capacitor-less EV charger with reduced component count and simple circuit structure. The proposed charger achieves ZVS turn-on of all switches and ZCS turn-off of all diodes without regard to voltage and load variation. It achieves high power density even without an input filter due to CCM operation and bulky electrolytic capacitors and without a low-frequency component in the transformer. A 2 kW prototype of the proposed charger with sinusoidal charging is built and tested to verify the validity of the proposed operation.

• Proceedings of the KIPE Conference
• /
• 2001.10a
• /
• pp.193-197
• /
• 2001

This paper presents a novel circuit topology of the double two-switch forward type high frequency transformer linked soft-switching PWM DC-DC power converter with tapped inductor filters that can operate under a condition of the low peak voltage stress across the power semiconductor devices and lowered peak current stress through the transformer for some high power applications. This circuit topology of an interleaved two-switch forward soft-switching power converter is proposed in the order to minimize an idle circulating current due to the tapped inductor filter without of any additional active auxiliary resonant-assisted snubber circuits, such as active resonant DC link snubbers and AC link snubbers, active resonant commutation leg link snubbers. The unique advantages of this power converter are less power circuit components and power semiconductor devices, constant frequency PWM scheme, cost effective configuration and wider soft-switching PWM operation range under PWM power regulations load variations. The practical effectiveness of the proposed soft-switching converter circuit topology is tested by simulations and is proved by experimental results received from the 500W-100kHz breadboard setup.

• Phonetics and Speech Sciences
• /
• v.12 no.3
• /
• pp.55-63
• /
• 2020

Voice conversion can be applied to various voice processing applications. It can also play an important role in data augmentation for speech recognition. The conventional method uses the architecture of voice conversion with speech synthesis, with Mel filter bank as the main parameter. Mel filter bank is well-suited for quick computation of neural networks but cannot be converted into a high-quality waveform without the aid of a vocoder. Further, it is not effective in terms of obtaining data for speech recognition. In this paper, we focus on performing voice-to-voice conversion using only the raw spectrum. We propose a deep learning model based on the transformer network, which quickly learns the voice conversion properties using an attention mechanism between source and target spectral components. The experiments were performed on TIDIGITS data, a series of numbers spoken by an English speaker. The conversion voices were evaluated for naturalness and similarity using mean opinion score (MOS) obtained from 30 participants. Our final results yielded 3.52±0.22 for naturalness and 3.89±0.19 for similarity.

• The Korean Journal of Ceramics
• /
• v.5 no.2
• /
• pp.171-177
• /
• 1999

In the $(Pb_{1-x}M_x)[(Mn_{1/3}Sb_{2/3})_{0.05}Zr_yTi_{0.95-y}]O_3$ system, where M=Ca and Sr, the piezoelectric properties were evaluated to examine the possibility of application to piezoelectric transformer. A Rosen-type piezoelectric transformer was formed, then the electrical properties of voltage step-up ratio, frequency characteristics etc. were analysed. The morphotropic phase boundary was determined to be y=0.475 in $Pb[(Mn_{1/3}Sb_{2/3})_{0.05}Zr_yTi_{0.95-y}]O_3$ system and the piezoelectric properties of this composition was kp=0.59, Qm=1600 and $\varepsilon_r$=1150. Moreover, when 1-2 mol% of Sr are substituted, enhanced piezoelectric properties of kp=0.61, Qm=1600 and $\varepsilon_r$=1400 were shown. The temperature rising (ΔT) of a piezoelectric transformer with $Pb[Mn_{1/3}Sb_{2/3})_{0.05}Zr_{0.475}Ti_{0.475})]O_3 $ composition was $10^{\circ}C$, and the voltage step-up ratio was 500 when the output voltage was 4000V, whereas the ΔT was below $3^{\circ}C$ and the resonant frequency variation ($\Delta f_r$) as a function of load resistance was below 5% when the output voltage was 2000 V. These characteristics are superior to the properties of materials, which were substituted by Ca or without substitution.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.17 no.1
• /
• pp.54-60
• /
• 2003

Feeding traction loads from the public power system may lead to some current and voltage unbalance and consequently affects the operation of its energy-supply system and other equipment connected with it. This paper introduces an analysis of the current unbalance caused by the demands of an electric railway on a public power system. And the results with compensator and without compensator are simulated, and eventually the formula about the current unbalance is suggested. The Scott-connected transformer is adopted in Korea National Railway System. So Scott-connected transformer among the various transformer connection schemes is analyzed in this paper. Also, the formulas about the unbalance and compensating current can be derived by using two parameters(M-phase and T-phase current) of secondary Scott-connected transformer. So, the practical and accurate simulation can be done through dynamic models by using scheduling of traction.

• Asia-pacific Journal of Multimedia Services Convergent with Art, Humanities, and Sociology
• /
• v.8 no.3
• /
• pp.563-570
• /
• 2018

This paper proposes an efficient dissolved gas analysis(DGA) and classification method of an oil-filled transformer using machine learning algorithms to solve problems inherent in IEC 60599. In IEC 60599, a certain diagnosis criteria do not exist, and duplication area is existed. Thus, it is difficult to make a decision without any experts since the IEC 60599 standard can not support analysis and classification of gas date of a power transformer in that criteria. To address these issue. we propose a dissolved gas analysis(DGA) and classification method using a machine learning algorithm. We evaluate the performance of the proposed method using support vector machines with dissolved gas dataset extracted from a power transformer in the real industry. To validate the performance of the proposed method, we compares the proposed method with the IEC 60599 standard. Experimental results show that the proposed method outperforms the IEC 60599 in the classification accuracy.

• Proceedings of the KIPE Conference
• /
• 2003.07a
• /
• pp.227-230
• /
• 2003

In this paper, a full bridge edge-resonant zero voltage mode based soft-switching PWM DC-DC power converter with a high frequency center tapped transformer link stage is presented from a practical point of view. The power MOSFETS operating as synchronous rectifier devices are implemented in the rectifier center tapped stage to reduce conduction power losses and also to extend the transformer primary side power MOSFETS ZVS commutation area from the rated to zero-load without a requirement of a magnetizing current. The steady-state operation of this phase-shift PWM controlled power converter is described in comparison with a conventional ZVS phase-shift PWM DC-DC converter using the diodes rectifier. Moreover, the experimental results of the switching power losses analysis are evaluated and discussed in this paper. The practical effectiveness of the ZVS phase-shift PWM DC-DC power converter treated here is actually proved by using 2.5kW-32kHz breadboard circuit. An actual efficiency of this converter is estimated in experiment and is achieved as 97$\%$ at maximum.

• Journal of Power Electronics
• /
• v.18 no.5
• /
• pp.1536-1544
• /
• 2018

The existing 24-pulse autotransformer rectifier unit (ATRU) needs interphase reactors for parallel work of the rectifier bridges, and its output voltage cannot be regulated. Aiming at these problems, a step-up and step-down asymmetrical 24-pulse ATRU is proposed in this paper. The connections and turns ratios among transformer windings are well designed. In addition, a 15-degree phase difference is formed between two of the 24 voltage vectors produced by the transformer, which makes the four rectifier bridge groups produce a 24-pulse DC voltage without interphase reactors. Meanwhile, by adding extended winding to each phase of the transformer, wide-range regulation of the ATRU output voltage can be realized, and the reasonable voltage regulation range is between 0.2 and 1.6. The superposition of the voltage vectors and the principle of the voltage regulation are analyzed in detail. Furthermore, the turns ratio of the windings, winding current, output voltage, and kilovolt-ampere rating are all derived. Finally, the simulations and experiments are carried out, and the correctness of the principle and theoretical analysis of the new 24-pulse ATRU are verified.