https://doi.org/10.5370/KIEE.2026.75.1.1

이해성(Haesung Lee) ; 정영범(Youngbeom Jung) ; 김우용(Wooyong Kim)

Accurate forecasting of load reductions during Demand Response (DR) events is essential for effective grid operation and settlement. This paper presents a context-aware deep learning model for predicting DR reductions using Advanced Metering Infrastructure (AMI) data, enhanced with contextual features such as weather, time, customer attributes, and DR history. The proposed model integrates a time-series encoder with multi-attention mechanisms, enabling dynamic weighting of input variables based on event-specific context. Experiments conducted on real-world AMI datasets demonstrate that the model outperforms baseline methods including linear regression, LSTM, and XGBoost in terms of MAE and RMSE. The results show improved prediction accuracy across customer types and DR scenarios. This approach enhances the interpretability and reliability of DR forecasting and provides a foundation for adaptive, context-sensitive DR management strategies. The model also supports applications in pre-settlement estimation and real-time DR operations.

https://doi.org/10.5370/KIEE.2026.75.1.10

맹종호(Jong-Ho Maeng) ; 김우중(Woo-Jung Kim) ; 이유석(Yu-Seok Lee) ; 전영환(Yeong-Han Chun)

This paper calculated the DLR (Dynamic Line Rating) using hourly temperature and wind speed data measured at meteorological observation stations. Inverse Distance Weighting (IDW) was applied to interpolate the meteorological data measured at weather stations to values for actual transmission lines. Additionally, for wind speed, the Terrain Roughness Index (TRI) was applied to quantify the surface roughness varying with terrain and improve the accuracy of the estimated wind speed. The accuracy of this methodology is evaluated using RMSE (Root Mean Square Error). Applying data from 10 meteorological stations yielded an RMSE of 1.24 to 1.33 for temperature and 0.78 to 0.79 for wind speed. The DLR calculation showed a 31.2% margin relative to the ACSR conductor's rated current during the normal condition in August, when temperatures were highest. Even during a single line fault on the 345kV line, a 6.5% margin was maintained.

https://doi.org/10.5370/KIEE.2026.75.1.19

최성문(Sung-Moon Choi) ; 유현상(Hyun-Sang You) ; 노성은(Seong-Eun Rho) ; 이중선(Joong-Seon Lee) ; 이승호(Seung-Ho Lee) ; 노대석(Dae-Seok Rho)

As the basic road-map for national carbon neutralization by 2030 and the 10th basic plan for long-term electricity supply and demand plan in Korea, the installation of renewable energy sources has been rapidly increased. However, the customer voltage has frequently deviated the allowable voltage limit (220V ± 6%) during periods with off peak load due to the large scaled installation of renewable energy sources. In order to solve these problems, this paper implements 30kW scaled VPL device, which is a virtual power line technology to improve hosting capacity of renewable energy sources instead of additional investment to power system infrastructure such as distribution substation and primary feeders and also proposes the optimal capacity for the VPL device in the secondary feeder. Furthermore, this paper performs modeling of VPL device in the secondary feeder, which is composed of battery system and operation platform, by using PSCAD/EMTDC S/W. From the simulation and test results based on the proposed VPL device in the secondary feeder, it is confirmed that it can keep the customer voltages within the allowable voltage limit, and then improve the hosting capacity of renewable energy sources efficiently without additional investment to power system infrastructure.

https://doi.org/10.5370/KIEE.2026.75.1.29

고병선(Byoung-Sun Ko) ; 차대석(Dae-Seak Cha) ; 채상헌(Sang Heon Chae) ; 최정식(Jung-Sik Choi)

This paper presents the development of a test evaluation system and the establishment of test procedures for assessing the performance of sub-modules applied to MVDC(medium voltage direct current) to MVDC converter stations. The proposed test system consists of a back to back circuit comprising four sub-modules and a power conversion device that supplies the operating voltage. In the absence of established standards and guidelines for MVDC systems, the test procedures are formulated based on the electrical test regulations for VSC(voltage source converter) valves used in HVDC(high voltage direct current) systems. Using the proposed test evaluation system, performance assessments of sub-modules for MVDC to MVDC converter stations are conducted.

https://doi.org/10.5370/KIEE.2026.75.1.36

이진수(Jinsoo Lee) ; 심상우(Sangwoo Shim) ; 노재형(Jae Hyung Roh) ; 박종배(Jong-Bae Park)

High PV penetration increases day-ahead scheduling risk by amplifying peak and ramp errors. We examine how spatial interpolation strength and loss-function choice affect short-term PV forecasts. Using Jeju Island data (plant outputs, 2020-01 ? 2024-05; ASOS stations Jeju and Gosan), we estimate plant-level meteorology via inverse distance weighting with exponent p∈{1,…,5} plus a spatial-average baseline, and train XGBoost models under three regression losses like MSE, MAE, and Pseudo-Huber. Training is restricted to daylight hours, and performance is evaluated using normalized mean absolute error. Results reveal strong seasonal interactions between interpolation strength and loss: the best seasonal nMAE is 5.48% (spring, IDW p=4 and MAE), 7.12% (summer, Average and Pseudo-Huber), 6.32% (autumn, IDW p=1 and Pseudo-Huber), and 5.42% (winter, IDW p=4 and MAE). These findings indicate that season-aware configuration of IDW and robust losses enhances peak/ramp tracking and reduces nMAE under sparse weather networks.

https://doi.org/10.5370/KIEE.2026.75.1.45

권태화(Tae-Hwa Kwon) ; 조창두(Chang-Doo Cho) ; 조창삼(Chang-Sam Cho) ; 정병창(Byoung-Chang Jeong)

Recently, the penetration of inverter-based renewable energy in the power grid is increasing rapidly. The proportion of large synchronous generators in the power grid is decreasing due to the increasing penetration of inverter-based renewable energy. As the number of large synchronous generators decreases, inverter-based renewable energy is increasing the need to supply transient current supplied by synchronous generators in the event of an accident in the power system, and to control the output reliably in weak power systems, GFM is emerging instead of GFL. In this paper, the inverter design elements are reviewed to supply the transient current required by the power grid and an improved GFM controller is proposed to limit the transient current.

https://doi.org/10.5370/KIEE.2026.75.1.54

김홍균(Hong-Gyun Kim) ; 최승수(Seung-Su Choi) ; 박영호(Young-Ho Park) ; 이신원(Shin-Won Lee) ; 임성훈(Sung-Hun Lim)

As global environmental concerns intensify, the development and advancement of renewable energy applications have gained increasing importance. However, the growing integration of distributed generation (DG) from renewable energy sources has amplified its impact on power system faults, prompting active research into protection strategies for power systems with DG. The increasing adoption of DG from renewable energy sources has resulted in a rise in fault current levels. To address this issue, superconducting fault current limiter (SFCL) has been introduced, leading to extensive research and ongoing commercialization. Another challenge of DG is the occurrence of overvoltage during islanding conditions. SFCL can contribute to voltage drops when an overvoltage occurs. To mitigate this problem, it is essential to analyze the protection coordination characteristics between the overcurrent relay (OCR) and the overvoltage relay (OVR). This study used PSCAD/EMTDC simulations to investigate the operation of the OCR, OVR, and magnetic-coupling type SFCL in mitigating overvoltage issues under islanding conditions in distributed generation. The results indicate that the application of the SFCL limits the current and causes a voltage drop, which in turn delays the operation of the OVR. This problem could be solved by adjusting the Time Dial(TD) setting of the OVR.

https://doi.org/10.5370/KIEE.2026.75.1.65

박종훈(Jong-Hoon Park) ; 한지훈(Ji-Hoon Han) ; 홍선기(Sun-Ki Hong)

In conventional hysteresis motors, finite element analysis has been challenging due to the unique characteristics of the rotor, and equivalent-circuit-based analysis has been the common approach. However, this method has the limitations in analyzing torque ripple caused by the stator slot structure. In this study, advanced finite element analysis techniques were applied to compare the torque characteristics and torque ripple with respect to the bridge angle of a closed-slot stator, and the noise characteristics were evaluated through prototype experiments. This approach enabled detailed analysis of the torque ripple characteristics of closed-slot hysteresis motors, which had been difficult to evaluate using conventional methods.

https://doi.org/10.5370/KIEE.2026.75.1.72

한수지(Suji Han) ; 주창대(Chan-Dae Joo) ; 천효석(Hyo-Seok Cheon) ; 김태규(Tae-Kue Kim)

In this paper, we presented a sensorless control method for a reciprocating compressor motor based on Single-Shunt current measurement. In order to mitigate the estimation performance degradation caused by current reconstruction errors, a dual algorithm-based sensorless transition method is employed. This method comprises a current model-based method at low speeds and an extended EMF observer at high speeds. Additionally, a resonant and repetitive controller has been integrated to suppress speed ripple caused by periodic load torque. The simulation results demonstrate the efficiency of the proposed approach in reducing both speed ripple and maximum estimation error across the entire speed range. These results is indicated that is a suitable approach for implementation in low-cost sensorless motor drive systems.

https://doi.org/10.5370/KIEE.2026.75.1.82

성철용(Chul-Yong Sung) ; 최형진(Hyung-Jin Choe)

This paper proposes a passive snubber circuit to reduce switching losses in boost converters. The proposed circuit integrates an inductor, two capacitors, and three diodes into the conventional boost converter topology, enabling simultaneous reduction of turn-on and turn-off switching losses. Experimental validation under conditions of input voltage of 24 V, output voltage of 100 V, switching frequency of 350 kHz, and output power ranging from 10 to 100 W demonstrates that the proposed circuit achieves a maximum efficiency improvement of 14.19% at 10 W and a minimum improvement of 1.34% at 100 W compared to the conventional boost converter. Furthermore, thermal imaging results demonstrate that the switch operating temperature of the proposed circuit is up to 40°C lower than that of the conventional circuit, even without a heatsink. These results confirm that the proposed passive snubber effectively enhances efficiency and thermal performance, and can contribute to the development of high-efficiency, high-reliability boost converters.

https://doi.org/10.5370/KIEE.2026.75.1.92

김지수(Ji-Su Kim) ; 정택근(Taek-Keun Jung) ; 김종수(Jong-Soo Kim)

This paper presents an integrated comparative analysis of resonant inductor core materials suitable for automotive power conversion systems, focusing on Ferrite (Mn-Zc, N87), High-Flux (Fe-Ni), and Sendust (Fe-Si-Al), based on combinations of their material properties, including eddy-current loss, hysteresis loss characteristics, and thermal behavior. Using Ansys Maxwell simulations, the loss behavior corresponding to the internal material property combinations of the magnetic cores is analyzed, and a resonant inductor core material with overall superior characteristics-such as mechanical strength and temperature performance-is selected. To validate the effectiveness of the proposed core selection method, a phase-shift full-bridge converter prototype and resonant inductors fabricated with the three materials are tested at a power level of 1. As a result, among the three experimentally evaluated materials, the system applying the High-Flux resonant inductor exhibits a total loss of 111 under 1 output conditions, while demonstrating superior mechanical robustness.

https://doi.org/10.5370/KIEE.2026.75.1.99

조예빈(Yebean Cho) ; (Sanh Vo Thi) ; (MD Arifur Rahman Barno) ; (Sourov Hossain) ; (Malkeshkumar Patel) ; 김준동(Joondong Kim)

Transparent ZnO/NiO heterojunction photodetectors based on the pyroelectric effect are demonstrated as promising devices that combine optical and thermal responses. Transparent photovoltaic (TPV) devices have attracted considerable attention as a next-generation energy technology that can be applied to windows of buildings and vehicles to generate electricity without requiring additional installation space. In this study, a transparent heterojunction structure (FTO/ZnO/NiO/AgNWs/ZnO) was fabricated to investigate the influence of the pyroelectric effect on photodetection performance, with the photovoltaic characteristics enabling self-operation. The device exhibited an average visible transmittance (AVT) of approximately 60.3%, and it showed clear photovoltaic operation with a relatively high photo-generated voltage (226 mV) and current density (92 A/cm2). This transparent photovoltaic device is possible to be applied for photodetector. This transparent photodetector provides the significantly high responsivity (19.511 A/W) and detectivity (1.8301013 Jones) due to the pyroelectric effect, and it maintains high values even at low frequencies, thereby operating effectively over a wide frequency range. These results indicate that the transparent heterojunction (p-NiO/n-ZnO) device is a photovoltaic cell to provide the operation power on photodetection, and thus next-generation photoelectric devices and applications.

https://doi.org/10.5370/KIEE.2026.75.1.107

차영광(Young-Kwang Cha) ; 이일회(Il-Hoe Lee) ; 전기범(Ki-Beom Jeon) ; 주흥진(Heung-Jin Ju)

A variable vacuum capacitor(VVC) is mainly used for impedance matching in semiconductor wafer and display panel manufacturing, where plasma load is required, due to its high dielectric strength and stability. As usage in these fields steadily increases, very long service life and reliability are required for VVCs. Among the internal components of the VVC, the bellows is manufactured by bending a thin metal sheet, making it structurally the most vulnerable part in terms of lifetime. Because a mechanical endurance of over 100,000 cycles is required, the bellows’ lifetime usually determines the lifetime of VVCs. In particular, the bellows is responsible not only for operation and maintaining vacuum tightness, but also for current carrying. Therefore, materials with high thermal conductivity are used for carrying high-frequency currents, or plating is applied to STS316L material. In this study, the mechanical endurance of the bellows used in VVCs was first tested under various plating conditions. After designing, analyzing, and manufacturing the bellows, they were plated with different solutions, and mechanical endurance tests were conducted. It was confirmed that the bellows showed the longest service lifetime when Au- plating was applied. However, due to increased manufacturing costs and management issues with Au-plating, VVCs were manufactured using bellows plated with copper, which showed the next best durability. Finally the mechanical endurance test of VVCs confirmed that the target service life requirement was met.

https://doi.org/10.5370/KIEE.2026.75.1.114

박재덕(Jae-Deok Park) ; 이동호(Dong-Ho Lee) ; 박태식(Tae-Sik Park)

Power transformers play a crucial role in power transmission and substation systems, and a systematic and efficient inspection strategy is essential for stable operation. However, existing TBI methods only perform inspections at regular intervals, failing to adequately reflect actual equipment conditions. This can lead to over-maintenance of low-risk equipment and insufficient management of high-risk equipment, resulting in inefficient investment costs. Therefore, this paper proposes a risk-based inspection interval calculation strategy to address these issues. In this study, FMEA of the transformer was performed, and based on this, PoF and CoF were calculated to derive risk. The inspection interval was selected based on the risk of the power transformer. To verify the validity of the proposed RBI, simulations were conducted. The results confirmed that it is an effective inspection strategy that addresses the issues of existing TBI and ensures cost-effectiveness and reliability.

https://doi.org/10.5370/KIEE.2026.75.1.121

이윤재(Yun-Jae Lee) ; 장성욱(Sung-Uk Zhang)

This study investigates the allowable current and the transient thermal response of MVDC power cables, specifically ACSR and XLPE types, using the Finite Element Method. To reflect real operating conditions, annual environmental temperature data from Suwon, Korea, were applied as dynamic boundary conditions. Steady-state analysis was first conducted to determine the ampacity, with the conductor temperature limited to 90 ℃. A monthly fault scenario was then simulated to evaluate the cooling and reheating rates, which characterize the thermal response of each cable type. The results show that ampacity strongly depends on cable structure and material properties. Overhead ACSR cables exhibit faster cooling and reheating than underground XLPE cables, due to their distinct heat dissipation mechanisms. These findings highlight the importance of a dynamic rating approach that accounts for environmental variations, rather than relying solely on static values. Furthermore, the quantified thermal response rates provide essential data for developing overload protection and operational strategies. This study therefore offers fundamental insights for the reliable and efficient design of future MVDC distribution networks.

https://doi.org/10.5370/KIEE.2026.75.1.129

신희성(Hee-Sung Shin) ; 홍상범(Sang-Beom Hong)

The growing demand for energy and the expansion of public infrastructure have led to a continuous increase in electricity consumption in public facilities. Among them, street lighting is one of the major energy-intensive infrastructures due to its wide deployment and long operating hours. Conventional streetlight control methods, based on fixed time or ambient brightness, suffer from inefficiency as they consume unnecessary power even during periods of very low traffic. To address this issue, this study proposes a distributed streetlight control system that selectively activates lights along the vehicle’s travel path by detecting vehicle approach in advance and using LoRa wireless communication. The proposed system utilizes traffic detection data from the National Transport Information Center to analyze traffic variations and implements a packet-based communication structure between nodes with separated detection and control functions. Simulation results demonstrate that in road segments with high nighttime non-traffic ratios, the proposed system can achieve more than 50% power savings. These results highlight that the LoRa-based distributed lighting control approach is particularly effective in low-traffic scenarios and can contribute to the sustainable energy management of public infrastructure.

https://doi.org/10.5370/KIEE.2026.75.1.135

전세현(Se-Hyun Jeon) ; 김정수(Jung-Su Kim)

This paper presents a smooth and robust missile guidance law that enables a missile to intercept a target despite the presence of unknown inputs such as target maneuvers and disturbances like wind. To address the performance degradation of conventional Proportional Navigation Guidance (PNG) caused by external disturbances and rapid target maneuvering, the proposed design defines a corrected target angle based on the geometric relationship between the missile and target velocity vectors, incorporating the collision triangle concept derived from maritime navigation. By applying Lyapunov stability theory, the proposed design ensures that the error angle converges to zero and provides a guidance law that is robust to uncertainties without the need for a separate disturbance observer. To handle uncertainties arising from unknown target motions and wind effects during the Lyapunov stability derivation, the control input is selected to include a smooth hyperbolic tangent function. Simulation results verify that the proposed guidance law exhibits robust performance, including a significantly reduced miss distance (less than 1m) and stable trajectories, across various target speeds, uncertain maneuvers, and operating conditions.

https://doi.org/10.5370/KIEE.2026.75.1.144

조성진(Sungjin Cho)

Most research on obstacle avoidance of unmanned aerial vehicles (UAVs) deals with complex path-planning algorithms with multiple modes and high computational time. However, frequently switching modes degrade vehicle control performance and long-time path-planing may lower the effectiveness of UAVs. For fast computation and simple structure of avoiding a stationary obstacle and reaching a target, we propose initial heading control of a proportional navigation law. By deriving the closed-form solution of the proportional navigation law, we obtain a maximum curvature condition incorporated by initial heading. We control initial heading angle to avoid an obstacle at the maximum curvature and to reach a target. Moreover, a compensation scheme is developed to avoid an obstacle located away from maximum curvature. The proposed schemes are validated by numerical simulations.

https://doi.org/10.5370/KIEE.2026.75.1.151

김준형(Jun Hyeong Kim) ; 김동헌(Dong Hun Kim)

In this study, a novel method combining neural networks with a PID controller to implement a 3-axis stabilization platform was proposed. In the 3-axis stabilization platform, the three servo motor angles corresponding to the plate's roll and pitch angles could not be directly measured. Due to the nonlinear characteristics in actual implementation, directly using a controller was challenging. Therefore, in the proposed study, we used a neural network to learn the relationship between the three servo motor inputs and the roll and pitch angles of the 3-axis stabilization. The neural network provided the servo motor inputs based on the roll and pitch angles measured by sensors. To improve the accuracy and reliability of the data, the measured values were corrected using a Kalman filter, and after data collection, the data was preprocessed for neural network modeling. The trained neural network model was combined with a PID controller to control the three servo motors, maintaining the plate's horizontal position. Experimental results demonstrated the performance comparison across various neural network structures, confirming that the roll and pitch angles of the oscillating plate were controlled by the proposed neural network-based stabilization.

https://doi.org/10.5370/KIEE.2026.75.1.159

신재훈(Jae-Hun Shin) ; 조성호(Sung-Ho Jo) ; 조유진(Yoo-Jean Cho) ; 김동욱(Dong-Wook Kim)

Network separation policies in power systems enhance security but limit real-time data integration between FA and OA networks, reducing operational efficiency and delaying responses to outages and faults. Conventional file-based data exchange methods are insufficient to ensure real-time synchronization and data consistency. This paper proposes a real-time data transmission and synchronization framework for heterogeneous networks in a network-separated environment, ensuring data consistency and ordering. The proposed system extracts incremental changes from the source database in real time, transmits them securely, and validates, orders, and synchronizes the data at the receiver. The results demonstrate that the proposed approach enables reliable real-time data sharing under network separation constraints, thereby improving the efficiency and stability of power system operations.

https://doi.org/10.5370/KIEE.2026.75.1.166

박영욱(Younguk Park) ; 계슬아(Seula Kye) ; 이언석(Onseok Lee)

Nailfold capillaroscopy is an efficient, noninvasive diagnostic method for predicting various diseases. However, its conventional application is constrained by the need for specialized equipment, hospital visits, and the reliance on the subjective judgment of the examiner, which undermines its objectivity. In this study, we propose a mobile-based capillary analysis system integrating a 3D-printed smartphone microscope and a deep learning model. We evaluated six segmentation models using data collected from the developed device, and U-Net demonstrated the best performance with a Dice Similarity Coefficient of 96.10% and an Accuracy of 93.23%. Furthermore, we implemented a client-server architecture application that enables real-time visualization of segmented images on mobile systems. This study demonstrates the potential to enhance healthcare accessibility by reducing economic and temporal burdens and offering a user-friendly diagnostic environment, particularly in underserved areas.

https://doi.org/10.5370/KIEE.2026.75.1.174

이수연(Su-Yeon Lee) ; 윤종서(Jong-Seo Yoon) ; 이석희(Seok-Hui Lee) ; 이정은(Jung-Eun Lee) ; 이정환(Jeong-Whan Lee)

This study aims to verify the physiological validity and reliability of the cardiac dipole model by generating and analyzing 3-dimensional vectorcardiogram (VCG) signals based on the Frank Lead system. First, we utilized ECG datasets from 30 healthy controls from the PTB Diagnostic ECG Database. The data was preprocessed to remove noise and baseline drift, and QRS peaks were detected to segment individual cardiac cycles. To minimize inter-individual variation, the signals were normalized based on the isoelectric PR segment. Using the normalized Frank Lead data (vx, vy, vz), a population-averaged VCG model was constructed. This model was then mapped onto a torso model that reflects Korean anatomical dimensions. We generated standard limb-leads (Lead I, II, III) by performing a vector projection between the VCG data and the RA, LA, and LL electrode positions. The correlation of these generated signals with measured ECG signals was then evaluated. The results showed a very high correlation for Lead I (r = 0.940) and Lead II (r = 0.823). This confirms that the proposed dipole model accurately reflects real cardiac electrical activity. In contrast, Lead III exhibited a lower correlation, which we attribute to inter-individual anatomical variability, such as rib position and torso length. In conclusion, this study demonstrates the reliability of the dipole-based VCG model, particularly for Leads I and II. The findings suggest that personalized modeling is necessary to improve the accuracy of Lead III. Furthermore, building on this validated dipole framework, the model can be extended to incorporate cardioid-based mathematical expressions to generate physiologically meaningful synthetic ECG signals for P, QRS, and T waves. This approach holds great potential for applications in cardiology simulations, algorithm development, and digital twin modeling.

https://doi.org/10.5370/KIEE.2026.75.1.182

박해산(Hae San Park) ; 강정원(Jeong Won Kang)

Following the opening of the Gyeongwon Line section between Dongducheon and Yeoncheon, a total of six power outages occurred due to failures of surge arresters installed on the overhead catenary system. On-train testing and analysis identified excessive leakage current and harmonic distortion as the primary causes. As an interim measure, an RC-Bank was installed at the Yeoncheon SSP, and the main transformer at Uijeongbu SS was operated in parallel to mitigate the leakage current. Subsequently a Smart Power Quality Improvement Device (SPQ) was deployed and its performance was verified under single transformer operation through live testing. The results demonstrated significant voltage stabilization and harmonic suppression. This study presents a practical case of SPQ application for improving power quality in railway power supply systems.

https://doi.org/10.5370/KIEE.2026.75.1.193

신현창(Hyun-Chang Shin) ; 김성철(Seong-Cheol Kim)

This paper presents the design and implementation of an all-IP?based IoT sensor data acquisition device tailored for railway trackside environments. The proposed system is developed with three primary objectives: Integration, enabling seamless aggregation and standardized processing of heterogeneous sensor data; Scalability, ensuring flexible interoperability with diverse sensor types through a unified hardware and communication architecture; and Maintainability, supporting efficient remote monitoring, diagnostics, and control for field maintenance operations. The device enables the transition from conventional periodic inspection practices to a real-time condition-monitoring and predictive-maintenance framework. By supporting proactive fault detection and enhancing trackside safety management, it contributes to improved operational reliability and reduced maintenance costs. Furthermore, the results of this study establish a technological foundation for advanced railway applications?including digital-twin systems, data-driven analytics, and AI-based predictive maintenance?demonstrating the potential of the proposed device for future expansion within intelligent railway infrastructure management platforms.

https://doi.org/10.5370/KIEE.2026.75.1.204

조걸(Zhao Jie) ; 방준호(Junho Bang) ; 최철영(Chul-Young Choi) ; 선로빈(Robin Sun) ; 박소연(Soyeon Park)

The recognition of electrical accident images is of great significance, but due to factors such as strong image noise interference and complex structure, traditional deep learning s often face the challenges of overfitting and insufficient generalization. To solve the above problems, this paper proposes a lightweight heterogeneous knowledge distillation framework for the classification of small sample electrical cable melting images. The framework uses U-Net3+ as the teacher network and ResNet-18 as the student network, introduces a multi-scale intermediate feature alignment module to alleviate the problem of feature inconsistency between heterogeneous structures, designs a composite distillation loss function, and introduces a label smoothing strategy in the output layer to enhance the regularization effect. The model performance is improved by combining the Warm-up and cosine annealing learning rate adjustment strategies. A systematic empirical analysis is conducted on a small sample dataset of 117 electrical cable melting images. The results show that the proposed method is significantly better than the baseline model and the traditional distillation scheme.

https://doi.org/10.5370/KIEE.2026.75.1.217

하디(Nazari Mohammad Hadi) ; 방준호(Junho Bang) ; 최철영(Chul-Young Choi) ; 선로빈(Robin Sun) ; 김든찬(Deunchan Kim)

In this paper, we proposed two VGG16-based hybrid models VGG16 + Dense and VGG16 + LSTM for the classification of short-circuit images used in analyzing the causes of electrical fires, and conducted a comparative performance evaluation of these models. The proposed models were trained using a transfer learning approach on an augmented dataset composed of two classes: primary short-circuit traces and secondary short-circuit traces. The VGG16 + Dense model achieved an accuracy of 95%, while the VGG16 + LSTM model achieved a higher accuracy of 99%, demonstrating superior performance in capturing spatial dependencies. Data augmentation techniques such as rotation, zooming, shifting, and brightness adjustment were applied to improve classification accuracy. As a result, the VGG16 + Dense model proved advantageous in terms of simplicity and faster training speed, whereas the VGG16 + LSTM model showed better precision and recall, making it more suitable for high risk applications where classification accuracy is critical. These results demonstrate that the proposed hybrid models offer enhanced characteristics compared to conventional single models, and they can be selectively applied based on specific field conditions.

https://doi.org/10.5370/KIEE.2026.75.1.223

조동일(Dong-Il Cho) ; 문원식(Won-sik Moon) ; 남준혁(Jun-Hyuk Nam) ; 조윤진(Yun-Jin Cho) ; 한성호(Seong-Ho Han)

Power transformers are critical assets in electrical grids, and accurately predicting their remaining useful life (RUL) is essential for predictive maintenance strategies. This paper presents a novel hybrid CNN-LSTM approach for Dissolved Gas Analysis (DGA) data that prioritizes computational efficiency while maintaining high prediction accuracy. The approach combines CNN for extracting spatial patterns from multi-dimensional gas concentrations and LSTM for modeling temporal dependencies, integrated with Cox proportional hazards regression for probabilistic survival predictions. Using 30 years of DGA data, the proposed model achieved a C-index of 0.822, comparable to LSTM-only model while reducing training time by 89.3%. This 9.3× faster training speed makes the model highly suitable for industrial deployment where rapid model updates are essential. Optimization techniques including CBAM and mixed-precision training further enhanced efficiency.

https://doi.org/10.5370/KIEE.2026.75.1.231

김동확(Dong-whak Kim) ; 김광호(Kwang-ho Kim)

This study predicts district-heating demand at the customer level and the heat-loss rate of a combined heat and power (CHP) plant, and then proposes an optimal daily thermal-storage capacity to enable efficient storage operation. The objective is to reduce total heat production while jointly optimizing electricity and heat output. We analyze operational data from a CHP plant that supplies a city in Gangwon-do, Korea, focusing on the winter months of January 2023 and January 2024. Customer-side consumption and network heat loss are quantified using district-heating metering data. For January 2024, we determine the optimal thermal-storage capacity and optimize charge/discharge schedules to minimize day-to-day state-of-charge variation. By stabilizing internal pressure fluctuations, the proposed strategy lowers distribution losses and improves overall heat-production efficiency. Simulation results indicate that the method reduces heat consumption at the CHP plant and lowers production costs.

https://doi.org/10.5370/KIEE.2026.75.1.238

석기혁(Gi-Hyeok SEOK) ; 김성열(Sung-Yul Kim)

This paper proposes the introduction of an Integrated Energy System (IES) into the conventional power system to address the uncertainties and surplus power associated with the increasing share of renewable energy sources (RES). By incorporating IES, this paper aims to mitigate the uncertainty of renewable generation facilities such as photovoltaic (PV) and wind turbines (WT), and to reduce excess electricity generation. In this research, surplus power caused by the oversupply of RES is absorbed through Electric Vehicle (EV) station and Power-to-Gas (P2G) technologies utilizing hydrogen energy. In addition, the shortage of thermal energy demand is supplemented using Power-to-Heat (P2H) technology based on Electric Heat Pumps (EHP). The study formulates a cost-benefit optimization model to determine the optimal capacity of P2X facilities by applying Mixed Integer Linear Programming (MILP), considering both capital and operating costs of each component in the IES under various RES penetration scenarios. Based on the optimization of the objective function, this paper presents an annual optimal operation strategy for the integrated energy system that maximizes the overall system benefit.