https://doi.org/10.5370/KIEE.2026.75.6.1239

강현구(Hyun-Koo Kang)

The design of new distribution networks requires a balanced consideration of economic efficiency and supply reliability. This paper proposes a distribution network line planning framework that explicitly considers the trade-off between these two objectives. The planning problem is formulated as a multi-objective optimization model incorporating line investment and operation costs as well as load interruption costs under single-line contingencies, and an improved novel binary particle swarm optimization (INBPSO) algorithm is applied to handle discrete line connection decisions. Simulation results on a 21-bus distribution system show that prioritizing economic efficiency leads to a radial network structure, whereas increasing the importance of reliability results in looped or partially meshed configurations. These results demonstrate that the economic–reliability trade-off directly determines the topological characteristics of new distribution networks.

https://doi.org/10.5370/KIEE.2026.75.6.1246

김준혁(Jun-Hyeok Kim)

As renewable energy expands, selecting representative solar profiles is crucial for power system planning. However, conventional methods based on central tendencies often smooth out intermittency, leading to an optimism bias that overestimates generation availability. To address this, this paper proposes a conservative profile generation method utilizing a multi-year weighted library and asymmetric loss functions. The proposed algorithm explicitly penalizes the frequency of over-estimation errors to mitigate the risk of phantom energy estimation. Case studies using 2017?2022 data from Korean sites verify that the proposed method reduces optimistic errors and maintains characteristic profile shapes compared with conventional mean-based and clustering-based baselines. By effectively capturing the lower bound of generation under high-variability conditions, such as spring intermittency and summer monsoons, the proposed approach provides a robust baseline for securing grid reliability.

https://doi.org/10.5370/KIEE.2026.75.6.1254

박성하(Sung-Ha Park)

This study conducts a quantitative assessment of protective relay performance based on a 345kV transformer cable ground-fault event within a power plant system. Fault waveforms captured during the incident, along with relay Event Recorder data, were analyzed to characterize current and voltage behavior under both steady-state and fault conditions and to compute the magnitude and duration of the ground-fault current. The results validate the appropriateness of the existing relay parameter settings, confirm the reliability of their fault-clearing operations, and verify proper coordination among the associated protection devices. The findings provide technical insights that can support the refinement of protection settings and enhance the overall dependability of power plant protection schemes.

https://doi.org/10.5370/KIEE.2026.75.6.1264

정태욱(Tae-Uk Jung)

Stricter environmental regulations by the International Maritime Organization (IMO) have increased the demand for high-efficiency and eco-friendly marine propulsion systems. Rim-driven thrusters (RDTs), which integrate the motor rotor with the propeller, offer advantages such as reduced mechanical losses, low noise, and improved maneuverability compared to conventional shaft-driven systems. This study investigates the feasibility of a 1.5 MW-class bearingless rim-driven propulsion motor by designing a 30 kW scaled permanent magnet synchronous motor (PMSM) for experimental validation. The motor is developed under volume constraints while meeting voltage, torque, and speed requirements, considering the large air gap of bearingless structures. Pole-slot combinations are selected and optimized to minimize torque ripple and cogging torque. Electromagnetic analysis verifies performance and confirms the feasibility of the proposed design for high-efficiency marine propulsion applications.

https://doi.org/10.5370/KIEE.2026.75.6.1271

조세진(Se-Jin Jo) ; 황윤성(Yun-Seong Hwang) ; 이상민(Sang-Min Lee) ; 이병국(Byoung-Kuk Lee)

This paper presents a comparative analysis of power conversion topologies to reduce stages and improve efficiency in Peltier-driven household refrigerators. Conventional architectures typically employ a multi-stage structure consisting of a PFC stage, an LLC resonant converter, and a Buck converter. While providing flexible control, this approach increases power loss and component count. To address these limitations, four single-stage isolated topologies?Phase-Shifted Full Bridge (PSFB), Asymmetric LLC, Active-Clamp Forward (AC Forward), and Active-Clamp Flyback (AC Flyback)?are evaluated under identical conditions through simulation. The results indicate that the AC Forward topology provides balanced efficiency over the entire load range. Based on this, an AC Forward-based power conversion system is implemented. Experimental results confirm improved efficiency compared with the conventional LLC?Buck architecture while significantly reducing power components.

https://doi.org/10.5370/KIEE.2026.75.6.1282

임종헌(Jong-hun Lim) ; 홍성준(Seong-jun Hong) ; 김태형(Tae Hyeong Kim) ; 임제영(Je Yeong Lim) ; 이병국(Byoung-Kuk Lee)

In battery management systems (BMS), accurate state-of-charge (SOC) estimation is essential for reliable and efficient operation of lithium-ion batteries. Conventional current integration methods suffer from cumulative errors over long-term operation, while voltage-based methods are sensitive to polarization effects and measurement noise, leading to unstable correction behavior. To address these limitations, this paper proposes an adaptive hybrid SOC estimation method that combines current integration with voltage-based correction. The proposed method derives a voltage residual using an equivalent circuit model and converts it into SOC correction based on the OCV?SOC relationship. In addition, adaptive correction factors and a mode-switching strategy are employed to regulate the correction strength and selectively apply voltage-based correction depending on operating conditions. Experimental validation using a long-term dynamic current profile demonstrates that the proposed method effectively suppresses cumulative error and improves estimation accuracy and stability across the entire SOC range.

https://doi.org/10.5370/KIEE.2026.75.6.1291

임성현(Sung-Hyun Lim) ; 이수호(Su-Ho Lee) ; 나준영(Jun-Young Na) ; 황재은(Jae-Eun Hwang) ; 유재성(Jae-Sung Yoo) ; 박혜리(Herie Park)

This study analyzes piezoelectric devices operating in the thickness vibration mode by examining models with various diameter-to-thickness ratios. Theoretical calculations and COMSOL finite element simulations, and experimental measurements were conducted using Pb(Ni1/3Nb2/3)O3-Pb(Zr Ti)O3-based ceramic materials. The results demonstrate that the resonance and anti-resonance frequencies generally decrease as the diameter-to-thickness ratio increases. The close agreement between experimental measurements and simulation results validates the reliability of the simulation approach.

https://doi.org/10.5370/KIEE.2026.75.6.1297

김현섭(Jong-Pil Lim)

This paper investigates the effect of carbon-doped GaN (C-GaN) layer thickness on the dynamic on-resistance (dynamic Ron) of AlGaN/GaN-on-Si heterojunction field-effect transistors (HFETs) using TCAD simulation. Three structures with different C-GaN thicknesses of 0.6, 0.9, and 1.2 μm were compared while keeping the unintentionally doped GaN (U-GaN) channel and AlGaN barrier identical. Off-state stress was applied at VGS=-10 V and VDS=25~300 V for 1000 s, followed by on-state recovery at VGS=0 V and VDS=1 V. Dynamic Ron was extracted from the drain current at 1 ms after turn-on. The results show that dynamic Ron increases as the C-GaN thickness increases under both transport conditions considered in this study. Electric field analysis reveals that thicker C-GaN causes stronger electric field concentration in the channel and buffer regions after off-state stress, resulting in slower 2DEG recovery. In addition, the case with relatively larger vertical leakage shows reduced channel electric field and smaller dynamic Ron, indicating that partial charge neutralization can mitigate buffer-induced current collapse. These results suggest that the thickness of the C-GaN layer is an important design parameter for improving dynamic switching performance in GaN-on-Si HFETs.

https://doi.org/10.5370/KIEE.2026.75.6.1304

정태환(Tae-Hwan Jeong) ; 김연수(Yeon-Su Kim) ; 장유나(You-Na Jang) ; 안달(Dal Ahn)

This paper proposes a gysel power divider/combiner structure capable of achieving wideband operation and second harmonic suppression. The proposed structure incorporates a stub-based bandpass filter design. Furthermore, heat dissipation characteristics are improved through vias inserted to implement short-circuited stubs. By achieving impedance matching at the center frequency of 3GHz. the proposed structure realizes a wide bandwidth ranging from 2.5GHz to 3.5GHz. Additionally, the characteristics of the short-circuited stubs are utilized to suppress the second harmonic at 6GHz. Experimental results demonstrate a return loss of less than ?18dB and a power division level of approximately ?3.2dB within the operating band, while achieving harmonic suppression of less than ?35dB at 6GHz

https://doi.org/10.5370/KIEE.2026.75.6.1314

임선우(Seon-Woo Im) ; 오정식(Jeong-Sik Oh) ; 박재덕(Jae-Deok Park) ; 박태식(Tae-Sik Park)

Power equipment has a characteristic that its failure probability increases exponentially with its service life, so initial failures of power equipment are rare. However, existing maintenance methods result in cost inefficiencies because maintenance is applied to healthy equipment that does not require maintenance. To address this issue, research is being conducted on risk-based maintenance strategies that prioritize equipment based on risk, considering the failure probability and ripple effects of failures, to manage equipment cost-effectively and ensure stable system operation. Since the failure probability of equipment varies depending on the maintenance applied, this should be considered when developing a maintenance strategy. Therefore, this paper applies a Risk-Based Maintenance (RBM) approach to develop a cost-effective maintenance strategy and verifies the performance of the RBM-based maintenance strategy.

https://doi.org/10.5370/KIEE.2026.75.6.1324

서영덕(Yeoung-Duk Seo) ; 지용진(Yong-Jin Ji) ; 권현호(Hyun-Ho Kwon) ; 김명진(Myung-Chin Kim)

This study proposes a Consequence of Failure(CoF) model and an Advanced Asset Investment Planning(AIP) algorithm to apply an ISO 55000-based power system asset management framework to load systems. Through asset management projects targeting private plants, practical opinions on replacement strategies and on-site considerations were collected from customers and modeled into criticality indices. These were then integrated into the AIP algorithm to satisfy the system risk requirements of various stakeholders. In particular, for each facility with a high Probability of Failure(PoF), the types of failures were bifurcated and reflected in AIP scenarios using a failure propagation weight. This design enables the establishment of more reliable investment plans for the short- and mid-term budgeting of private plants. The Advanced AIP proposed in this study serves as a consistency model for load system asset management, proving that it incorporates additional power-reception-side considerations beyond traditional utility-centered asset management. Furthermore, it presents a practical methodology that can scientifically contribute to the business viability of asset management by satisfying business constraints even with limited budgets.

https://doi.org/10.5370/KIEE.2026.75.6.1332

김예슬(Ye-Sle Kim) ; 전태현(Tae-Hyun Jun) ; 곽병섭(Byeong-Sub Kwak) ; 김아름(Ah-Reum Kim) ; 임병훈(Byung-Hun Lim) ; 박현주(Hyun-Joo Park)

As the service years of power transformers increase, ensuring the insulation reliability of bushings has become critical for power system stability. Although the influx of main tank oil during maintenance can degrade bushing insulation, standardized permissible mixing limits remain undefined. In this study, internal residual oil volumes during sampling were quantified through dismantling and structural modeling of three decommissioned bushings. Based on these findings, the impact of oil mixture on bushing integrity was evaluated using accelerated aging tests. Experimental results demonstrated that while dielectric characteristics remained stable, the pour point and kinematic viscosity variation rate functioned as the primary limiting factors for operational reliability. Analysis of each parameter yielded conservative critical mixing ratios: 30% for pour-point-dominant oils, and 20% for oils governed by viscosity variation standards. Furthermore, this study confirmed that the viscosity variation rate exhibited a high correlation with a linear regression model, enabling precise property prediction based on the mixing ratio. Finally, a maintenance algorithm was proposed to determine replenishment limits and sampling availability in the field by integrating critical values with residual oil data. This study is expected to provide data-driven guidelines for systematic bushing oil mixture management, thereby enhancing the long-term reliability of transformer assets.

https://doi.org/10.5370/KIEE.2026.75.6.1343

김준영(Jun-young Kim) ; 박기쁨(Gi-Ppeum Park) ; 심성민(Sung-min Sim) ; 김정무(Jung-Mu Kim)

In this paper, we analyze the impedance characteristics of free-falling objects utilizing a cylindrical electrode sensor and propose a contactless material classification method. Steel, glass, and rubber beads were used as materials that freely fall into the cylindrical electrode sensor. Using a transimpedance amplifier (TIA), the current induced in the cylindrical electrode by impedance variations was converted into a voltage signal and measured. Using a lock-in amplifier-based (LIA-based) measurement system, we detected the impedance signal variations of the cylindrical electrode corresponding to the electromagnetic properties of materials. In measurement results, distinct signal patterns were observed: the steel bead exhibited a maximum amplitude change of 3.88 mV and a phase change of 1.00°, while glass and rubber beads showed changes of 2.75 mV/0.20° and 2.31 mV/0.14°, respectively. The measurements for glass beads of varying sizes demonstrated a clear correlation between the bead diameter and the signal magnitude. The amplitude and phase variations of 30 mm bead were 8.54 mV and 1.46°, while the 26 mm, 21 mm, and 16 mm beads exhibited sequentially decreasing amplitudes of 6.11 mV, 2.75 mV, and 1.07 mV, accompanied by phase changes of 1.17°, 0.21°, and 0.17°, respectively. The results are verified that the electromagnetic properties of the materials can be effectively distinguished by monitoring the simultaneous changes in the amplitude and phase of the impedance signal.

https://doi.org/10.5370/KIEE.2026.75.6.1352

김휘중(Hwi-Jung Kim) ; 박소윤(So-Yoon Park) ; 한영웅(Young-Woong Han) ; 배원규(Won-Gyu Bae)

The AI semiconductor market is projected to grow rapidly, but Korea’s non-memory sector still holds a low market share and relies heavily on foreign technologies. This paper analyzes patent trends in third- and fourth-generation AI semiconductor technologies to provide strategic insights for enhancing Korea's technological competitiveness. The study categorizes next-generation AI semiconductors into processing and hardware domains, focusing on a qualitative analysis of hardware based on quantitative comparisons. A classification table was developed, including core technologies such as neuromorphic, NPU, HBM, CXL, PIM, and advanced packaging. Patent data from Korea, the U.S., and China were used to evaluate technology concentration by company and country. The results identify focused and core technologies by leading applicants and offer implications for guiding Korea’s development strategies in the AI semiconductor sector.

https://doi.org/10.5370/KIEE.2026.75.6.1367

양찬희(Chan-Hee Yang) ; 오성권(Sung-Kwun Oh)

In this study, an Iterative Refined Radial Basis Function Neural Network (IRRBFNN) is proposed for regression tasks. The model leverages error information generated at each inference step to sequentially ⅰ) localize the input space via Weighted Fuzzy C-means(WFCM) and ⅱ) optimize connection weights through Weighted Least Square Estimation (WLSE). At this stage, to optimize the number of local regions, a high-error region search and splitting scheme was applied, and acceptance criteria were used to determine whether a split should be executed. When evaluated on 16 benchmark datasets against state-of-the-art (SOTA) baselines, the proposed method achieves superior performance on 10 datasets. A case study on cement strength prediction further demonstrates the advantages of the IRRBFNN over existing application-specific models, highlighting its practical applicability to forecasting the properties of materials and related regression problems.

https://doi.org/10.5370/KIEE.2026.75.6.1378

채운(Woon Chae) ; 서기성(Kisung Seo)

Cross-modality Person Re-Identification aims to identify the same individual across color images captured during daytime and infrared images captured at nighttime. Due to the significant visual discrepancy between these modalities, it is considered substantially more challenging than conventional person re-identification. To alleviate the modality gap, extensive studies have been conducted on representation learning, network architecture design, and loss function optimization. In this paper, we focus on data augmentation, which has received relatively limited attention but plays a crucial role in cross-modality learning. Existing augmentation methods primarily aim to increase diversity within a single modality, which limits their ability to adequately reflect the distribution discrepancy across different modalities. To address this limitation, we propose a data augmentation framework that integrates GAN-based modality transformation with mixed-modality information to simultaneously enhance data diversity and effectively reduce modality discrepancies. To validate the effectiveness of the proposed method, we conduct extensive experiments on SYSU-MM01, a widely used benchmark dataset for cross-modality person re-identification, and compare our approach with various state-of-the-art methods. Experimental results demonstrate the superiority of the proposed method.

https://doi.org/10.5370/KIEE.2026.75.6.1383

원형식(Hyeong-sik Won) ; 조현종(Hyun-chong Cho)

Timely detection of farrowing in sows is important for effective farm management and animal welfare. However, existing farrowing monitoring approaches are largely contact-based, which limits their practical applicability in farm environments due to constraints such as cumulative equipment costs and animal stress. To address these limitations, this study developed a deep learning-based non-contact farrowing classification system for sows. The proposed method employed precise region of interest (ROI) cropping based on the Segment Anything Model (SAM) to reduce background interference and consistently include farrowing-related regions. In addition, multiple-instance learning was integrated into a Convolutional Neural Network (CNN)-based classification framework to better aggregate region-wise discriminative cues. Experimental results showed that the proposed method achieved the best overall performance among the compared models, with 85.47% recall and 85.68% F1-score. Compared with the original full-image input, it improved recall by 3.59 percentage points and F1-score by 3.71 percentage points. These results indicate that precise ROI cropping and multiple-instance learning jointly improve farrowing classification performance by reducing background interference and better aggregating region-wise discriminative cues.

https://doi.org/10.5370/KIEE.2026.75.6.1390

김민중(Min-Joong Kim) ; 김현우(HyeonWoo Kim)

Real estate transaction data in Korea are inherently event-driven, resulting in irregular observation intervals and extended periods without transactions, which make it difficult to construct continuous time-series representations. To address this issue, this study proposes a volatility-aware reconstruction method that transforms fragmented transaction records into continuous apartment-level time-series. Transaction data are reorganized into monthly sequences, and unobserved intervals are reconstructed by jointly considering local temporal continuity and region-level price variations. Experimental results demonstrate that the proposed reconstruction-based approach consistently outperforms conventional strategies that discard incomplete observations, achieving superior predictive performance. Furthermore, incorporating regional market volatility leads to additional performance gains.

https://doi.org/10.5370/KIEE.2026.75.6.1400

김정현(Jeonghyeon Kim) ; 김한솔(Han Sol Kim) ; 이창은(Changeun Lee) ; 이광일(Kwangil Lee)

In this paper, we propose a dynamic fusion method that combines varifocal loss (VFL) and seesaw loss (SSL) to address the class imbalance problem in long-tail datasets for object detection models. The static combination of two loss functions makes the training unstable due to the noisy gradients of the SSL, which interrupt the IoU-aware classification flow and limit overall performance. To address this problem, we aim to effectively mitigate the biased learning problem in long-tailed datasets by maintaining the stability of IoU-aware classification during the early learning stages and gradually reflecting the calibration effect of the SSL in the latter stages of training. Finally, we validate unbiased detecting performance of the object detection on the LVIS dataset. To this end, we applied the proposed loss fusion method to the RT-DETRv2 model, resulting in 35.4% of bias mitigation for rare classes.

https://doi.org/10.5370/KIEE.2026.75.6.1406

이재원(Jae-Won Lee) ; 최진규(Jin-Kyu Choi) ; 유성현(Seong-Hyeon Ryu) ; 안세환(Se-Hwan An) ; 신영철(Young-Cheol Shin) ; 홍순일(Soon-Il Hong) ; 류제덕(Jae-Deok Ryu) ; 이지영(Ji-Young Lee) ; 민경민(Kyeong-Min Min) ; 주지한(Ji-Han Joo)

This paper describes the development of an RFSoC-based signal processing unit for application in a compact AESA tracking radar. In addition to conventional signal processing functions, the proposed system supports operational signal generation. Accordingly, the system is designed by dividing its functions into four main components: DAQ, SIG, OPC, and PWR. The DAQ receives five-channel input signals with a center frequency of 10MHz or higher and performs preprocessing. First, the input signals are sampled using the RFSoC integrated ADC, followed by frequency down-conversion. Additional digital down conversion is implemented within the FPGA internal logic. Subsequently, Window, FFT processing, and data type conversion are applied to process the signals. The processed data are then delivered to the DSP for post-processing. The SIG generates three-channel operational signals required for system operation. Each signal has a center frequency of 1GHz or higher and a bandwidth exceeding 100MHz or higher. These signals are generated through the RFSoC integrated DAC and FPGA internal logic and are output at a sampling rate of 2GSPS or higher. The OPC receives the preprocessed data from the DAQ and performs post-processing while configuring the required operational signals. Using DSPs, target parameters such as range, direction, and velocity are analyzed for target detection. The PWR supplies all required power to the DAQ, SIG, and OPC boards. This verified the development of a broadband multi-operation mode signal processing unit. Through this, the applicability to a Small AESA Tracking Radar was verified, and the feasibility of miniaturization was confirmed.

https://doi.org/10.5370/KIEE.2026.75.6.1413

이행섭(Haeng Seob Lee) ; 김경화(Kyeong-Hwa Kim)

To secure the safety and operational reliability of rail freight transportation, there is a growing need for technologies that can detect faults in key freight car components at an early stage and prevent failures in advance. Wheel cracks and axle bearing abnormalities can rapidly deteriorate and lead to catastrophic failures during operation, thereby increasing the risk of serious accidents. However, inspection frameworks centered on periodic maintenance alone have limitations in timely reflecting the occurrence and progression characteristics of such defects, which underscores the need to establish an in-service early detection scheme. In this study, signals generated from a theoretical bearing fault model and field data acquired during freight car operation were processed using the same frequency-domain fault-signal analysis pipeline. The reproducibility of defect-related characteristic frequency components and impulsive responses was then compared to assess field applicability.

https://doi.org/10.5370/KIEE.2026.75.6.1421

노강일(Kang-Il Ro) ; 함동령(Dong-Young Ham) ; 구자영(Ja-Young Koo) ; 곽준호(Jun-Ho Kwak) ; 김희동(Hee-Dong Kim)

Stator windings in hydro-generators undergo progressive degradation over long-term operation due to combined thermal, electrical, mechanical, and environmental stresses. These stresses facilitate the formation of internal voids within insulation materials, degradation of semi-conductive layers, and surface fouling of end-windings, which collectively induce phase-to-phase discharges and eventually result in catastrophic insulation failure. Recently, the rapid expansion of wind and solar power has intensified the intermittency of renewable energy sources. This shift has forced hydro-generators to undergo more frequent start-stop cycles, leading to the formation of defects at welded joints connecting the stator windings a phenomenon increasingly linked to recent insulation failures. This paper presents a root cause analysis (RCA) of stator winding insulation failures in two hydro-generators following long-term operation. A hydro-generator (26.76 MVA, 10.2 kV) in service for 14 years, experienced a phase-to-phase insulation breakdown between the U and V phases. Another hydro-generator (27.8 MVA, 11 kV) after 31 years of operation, exhibited abnormal thermal behavior, with U-phase temperatures peaking at 134°C, which necessitated an emergency shutdown. Based on the RCA, subsequent repairs and component replacements were performed. Finally, diagnostic insulation tests were conducted to verify the integrity and reliability of the restored stator windings.

https://doi.org/10.5370/KIEE.2026.75.6.1427

이준원(Junwon Lee)

With the widespread adoption of cloud infrastructure and remote work, encrypted remote access traffic via protocols such as SSH and RDP has surged, emerging as a primary attack vector of network intrusions. Attackers are bypassing traditional Deep Packet Inspection through non-standard port manipulation and payload encryption, necessitating AI-based detection technologies capable of analyzing behavioral patterns without traffic decryption. This paper systematically presents design guidelines for an encrypted traffic detection platform, examining key challenges and the technological evolution of detection models across three stages: data collection, preprocessing, and detection model design. In particular, the author's empirical research on ML-based detection models and WGAN-GP based synthetic data generation is presented as practical case studies, and future research directions for anomaly behavior detection are proposed. Furthermore, this study explores the potential of future LLM based traffic synthesis through experimental analysis. Notably, incorporating synthetic traffic synthesized by the LLM-based model achieved a significant 21.4% enhancement in F1-score performance.

https://doi.org/10.5370/KIEE.2026.75.6.1435

박병철(Byung-Chul Park)

Lithium-ion batteries are core components of electric vehicle (EV) and energy storage system (ESS). As their deployment expands, fire incidents have also increased, highlighting the growing importance of state diagnosis technologies. Electrochemical Impedance Spectroscopy (EIS) is a representative method for quantitatively evaluating the degradation state and electrochemical reactions of batteries by analyzing internal impedance characteristics across different frequencies. However, commercial EIS measurement equipment is designed for testing individual cells rather than real-time monitoring, making it difficult to implement in on-site battery systems for real-time data acquisition. In this paper, we propose a method for repeatedly acquiring online EIS data while integrated with the battery system, implement it into an actual device, and verify the validity of the EIS measurements. Furthermore, we present an artificial intelligence (AI) learning model to utilize the acquired EIS data for battery state classification. By using the measured EIS data as input, we demonstrate that state classification and evaluation are feasible through the proposed model.

https://doi.org/10.5370/KIEE.2026.75.6.1442

김수완(SuWan KIM) ; 이개명(Gae-Myoung Lee)

Rapid EV charger expansion has caused spatial mismatches. We analyzed idle rates of 23,769 Jeju EV chargers using 285,228 smart meter records (2024) using one 1 year accumulated data. Defining 'idle' as zero consumption for three consecutive months, chargers were classified into four tiers: Open-Fast, Open-Slow, Private-Fast, and Private-Slow. K-Means clustering and t-tests revealed significant disparities. Suburban 'Open-Slow' chargers showed critical idle rates due to maintenance abandonment and blind public allocation. 'Private-Fast' chargers exhibited extreme loads with zero idleness, while 'Private-Slow' units displayed sporadic idleness globally. Considering legal limits on removing private property, we propose a grid management framework: a negative list restricting new installations in high-idle zones, financial incentives for voluntary relocation, and automatic contract power suspension for idle chargers. This data-driven approach prevents budget waste and reclaims grid hosting capacity.