• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.22 no.3
• /
• pp.131-146
• /
• 2023

Smart card data is widely used in the public transportation field. Despite the inevitability of various errors occur during the data collection and storage; however, smart card data errors have not been extensively studied. This paper investigates inherent errors in boarding and alighting station information in smart card data. A comparison smart card data and bus boarding and alighting survey data for the same time frame shows that boarding station names differ by 6.2% between the two data sets. This indicates that the error rate of smart card data is 6.2% in terms of boarding station information, given that bus boarding and alighting survey data can be considered as ground truth. This paper propose 6-step algorithm for correcting errors in smart card boarding station information, linking them to corresponding information in Bus Information System(BIS) Data. Comparing BIS data and bus boarding and alighting survey data for the same time frame reveals that boarding station names correspond by 98.3% between the two data sets, indicating that BIS data can be used as reliable reference for ground truth. To evaluate its performance, applying the 6-step algorithm proposed in this paper to smart card data set shows that the error rate of boarding station information is reduced from 6.2% to 1.0%, resulting in a 5.2%p improvement in the accuracy of smart card data. It is expected that the proposed algorithm will enhance the process of adjusting bus routes and making decisions related to public transportation infrastructure investments.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2021.06a
• /
• pp.88-88
• /
• 2021

최근 수자원과 수질관리 분야에 자료기반 머신러닝 모델과 딥러닝 모델의 활용이 급증하고 있다. 그러나 딥러닝 모델은 Blackbox 모델의 특성상 고전적인 질량, 운동량, 에너지 보존법칙을 고려하지 않고, 데이터에 내재된 패턴과 관계를 해석하기 때문에 물리적 법칙을 만족하지 않는 예측결과를 가져올 수 있다. 또한, 딥러닝 모델의 예측 성능은 학습데이터의 양과 변수 선정에 크게 영향을 받는 모델이기 때문에 양질의 데이터가 제공되지 않으면 모델의 bias와 variation이 클 수 있으며 정확도 높은 예측이 어렵다. 최근 이러한 자료기반 모델링 방법의 단점을 보완하기 위해 프로세스 기반 수치모델과 딥러닝 모델을 결합하여 두 모델링 방법의 장점을 활용하는 연구가 활발히 진행되고 있다(Read et al., 2019). Process-Guided Deep Learning (PGDL) 방법은 물리적 법칙을 반영하여 딥러닝 모델을 훈련시킴으로써 순수한 딥러닝 모델의 물리적 법칙 결여성 문제를 해결할 수 있는 대안으로 활용되고 있다. PGDL 모델은 딥러닝 모델에 물리적인 법칙을 해석할 수 있는 추가변수를 도입하며, 딥러닝 모델의 매개변수 최적화 과정에서 Cost 함수에 물리적 법칙을 위반하는 경우 Penalty를 추가하는 알고리즘을 도입하여 물리적 보존법칙을 만족하도록 모델을 훈련시킨다. 본 연구의 목적은 대청호의 수심별 수온을 예측하기 위해 역학적 모델과 딥러닝 모델을 융합한 PGDL 모델을 개발하고 적용성을 평가하는데 있다. 역학적 모델은 2차원 횡방향 평균 수리·수질 모델인 CE-QUAL-W2을 사용하였으며, 대청호를 대상으로 2017년부터 2018년까지 총 2년간 수온과 에너지 수지를 모의하였다. 기상(기온, 이슬점온도, 풍향, 풍속, 운량), 수문(저수위, 유입·유출 유량), 수온자료를 수집하여 CE-QUAL-W2 모델을 구축하고 보정하였으며, 모델은 저수위 변화, 수온의 수심별 시계열 변동 특성을 적절하게 재현하였다. 또한, 동일기간 대청호 수심별 수온 예측을 위한 순환 신경망 모델인 LSTM(Long Short-Term Memory)을 개발하였으며, 종속변수는 수온계 체인을 통해 수집한 수심별 고빈도 수온 자료를 사용하고 독립 변수는 기온, 풍속, 상대습도, 강수량, 단파복사에너지, 장파복사에너지를 사용하였다. LSTM 모델의 매개변수 최적화는 지도학습을 통해 예측값과 실측값의 RMSE가 최소화 되로록 훈련하였다. PGDL 모델은 동일 기간 LSTM 모델과 동일 입력 자료를 사용하여 구축하였으며, 역학적 모델에서 얻은 에너지 수지를 만족하지 않는 경우 Cost Function에 Penalty를 추가하여 물리적 보존법칙을 만족하도록 훈련하고 수심별 수온 예측결과를 비교·분석하였다.

• Journal of the Korea Computer Graphics Society
• /
• v.29 no.3
• /
• pp.117-125
• /
• 2023

In the development of clothing design through virtual simulation, it is essential to minimize the differences between the virtual and the real world as much as possible. The most critical task to enhance the similarity between virtual and real garments is to find simulation parameters that can closely emulate the physical properties of the actual fabric in use. The simulation parameter optimization process requires manual tuning by experts, demanding high expertise and a significant amount of time. Especially, considerable time is consumed in repeatedly running simulations to check the results of applying the tuned simulation parameters. Recently, to tackle this issue, artificial neural network learning models have been proposed that swiftly estimate the results of drape test simulations, which are predominantly used for parameter tuning. In these earlier studies, relatively simple linear stiffness models were used, and instead of estimating the entirety of the drape mesh, they estimated only a portion of the mesh and interpolated the rest. However, there is still a scarcity of research on non-linear stiffness models, which are commonly used in actual garment design. In this paper, we propose a learning model for estimating the results of drape simulations for non-linear stiffness models. Our learning model estimates the full high-resolution mesh model of drape. To validate the performance of the proposed method, experiments were conducted using three different drape test methods, demonstrating high accuracy in estimation.

• The Journal of Bigdata
• /
• v.7 no.2
• /
• pp.31-41
• /
• 2022

Steel is one of the most fundamental components to mechanical industry. However, the quality of products are greatly impacted by the surface defects in the steel. Thus, researchers pay attention to the need for surface defects detector and the deep learning methods are the current trend of object detector. There are still limitations and rooms for improvements, for example, related works focus on developing the models but don't take into account real-time application with practical implication on industrial settings. In this paper, a real-time application of steel surface defects detection based on YOLOv4 is proposed. Firstly, as the aim of this work to deploying model on real-time application, we studied related works on this field, particularly focusing on one-stage detector and YOLO algorithm, which is one of the most famous algorithm for real-time object detectors. Secondly, using pre-trained Yolov4-Darknet platform models and transfer learning, we trained and test on the hot rolled steel defects open-source dataset NEU-DET. In our study, we applied our application with 4 types of typical defects of a steel surface, namely patches, pitted surface, inclusion and scratches. Thirdly, we evaluated YOLOv4 trained model real-time performance to deploying our system with accuracy of 87.1 % mAP@0.5 and over 60 fps with GPU processing.

• Journal of Korea Water Resources Association
• /
• v.56 no.5
• /
• pp.337-345
• /
• 2023

Monitoring river microplastics is a challenging task since it is a time-consuming and high-cost process. The use of a physical model to have a better understanding of river microplastics' behaviors can complement the challenging monitoring process. However, there have been very limited studies on modeling river microplastics. In this study, therefore, we evaluated the applicability of one commonly used river water quality model, i.e., the Water Quality Analysis Simulation Program (WASP), in simulating the microplastic concentration in the river environment. We simulated the microplastic concentration in the Anyangcheon stream using the WASP's biochemical oxygen demand (BOD) and suspended solid (SS) variables as possible surrogate variables for the microplastics. Simulation analyses indicate that the SS state variable performs better than the BOD state variable to mimic the observed concentrations of microplastics. This is because of the characteristics of each water quality parameter; the BOD variable, a biochemical indicator, is inappropriate for modeling the behaviors of microplastics, which have generally constant biochemical features. In contrast, the SS variable, which has similar physical behaviors, followed the observed patterns of the microplastic concentrations well. To build a more advanced and accurate model for simulating the microplastic concentration, comprehensive and long-term monitoring studies of the river microplastics under different environmental conditions are needed, and the unit of microplastic concentration should be carefully addressed before its modeling application.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.4
• /
• pp.21-29
• /
• 2023

In this paper, we propose an AI-based system for automatically classifying industry and occupation codes in the population census. The accurate classification of industry and occupation codes is crucial for informing policy decisions, allocating resources, and conducting research. However, this task has traditionally been performed by human coders, which is time-consuming, resource-intensive, and prone to errors. Our system represents a significant improvement over the existing rule-based system used by the statistics agency, which relies on user-entered data for code classification. In this paper, we trained and evaluated several models, and developed an ensemble model that achieved an 86.76% match accuracy in industry and 81.84% in occupation, outperforming the best individual model. Additionally, we propose process improvement work based on the classification probability results of the model. Our proposed method utilizes an ensemble model that combines transfer learning techniques with pre-trained models. In this paper, we demonstrate the potential for AI-based systems to improve the accuracy and efficiency of population census data classification. By automating this process with AI, we can achieve more accurate and consistent results while reducing the workload on agency staff.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.28 no.2A
• /
• pp.259-267
• /
• 2008

The ultimate strengths of reinforced concrete deep beams are governed by the capacity of the shear resistance mechanism composed of concrete and shear reinforcing bars, and the structural behaviors of the beams are mainly controlled by the mechanical relationships according to the shear span-to-effective depth ratio, flexural reinforcement ratio, load and support conditions, and material properties. In this study, a simple indeterminate strut-tie model reflecting all characteristics of the ultimate strengths and complicated structural behaviors is presented for the design of simply supported reinforced concrete deep beams. In addition, a load distribution ratio, defined as a magnitude of load transferred by a vertical truss mechanism, is proposed to help structural designers perform the design of simply supported reinforced concrete deep beams by using the strut-tie model approaches of current design codes. In the determination of a load distribution ratio, a concept of balanced shear reinforcement ratio requiring a simultaneous failure of inclined concrete strut and vertical steel tie is introduced to ensure the ductile shear failure of reinforced concrete deep beams, and the prime design variables including the shear span-to-effective depth ratio, flexural reinforcement ratio, and compressive strength of concrete influencing the ultimate strength and behavior are reflected upon based on various and numerous numerical analysis results. In the companion paper, the validity of presented model and load distribution ratio was examined by employing them to the evaluation of the ultimate strengths of various simply supported reinforced concrete deep beams tested to failure.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.28 no.2A
• /
• pp.297-304
• /
• 2008

It is well known that cracks in concrete decrease permeability and durability of concrete because cracks enhance the penetration of water or corrosive chemicals like as chlorides, carbon dioxides, sulfates and some others. But some of cracks in hardened cements may be sealed in case of contacting water. This phenomenon is called "self healing" and it has a close relation to hydration products newly formed on surfaces of cracks. Many studies on self healing in concretes commonly showed that CSH gel has been observed on crack surfaces. And some studies have reported that calcium hydroxides and ettringite were observed as well as CSH gel on crack surfaces. This study was carried out to investigate hydration products formed by self healing process and also examine the influence of waterproof admixture for concretes on self healing of cement. As a result of XRD, DSC, SEM and EDX analysis of crack surfaces, it was found that self healing of cement was related to CSH gel, calcium hydroxides and ettringite. And waterproof admixture increased fibrous (needle-like) hydration products which were in network form. It is estimated that such fibrous products are effective for self healing process of cement system.

• KIPS Transactions on Software and Data Engineering
• /
• v.12 no.7
• /
• pp.303-314
• /
• 2023

Voice conversion, a technology that allows an individual's speech data to be regenerated with the acoustic properties(tone, cadence, gender) of another, has countless applications in education, communication, and entertainment. This paper proposes an approach based on the StarGAN-VC model that generates realistic-sounding speech without requiring parallel utterances. To overcome the constraints of the existing StarGAN-VC model that utilizes one-hot vectors of original and target speaker information, this paper extracts feature vectors of target speakers using a pre-trained version of Rawnet3. This results in a latent space where voice conversion can be performed without direct speaker-to-speaker mappings, enabling an any-to-any structure. In addition to the loss terms used in the original StarGAN-VC model, Wasserstein distance is used as a loss term to ensure that generated voice segments match the acoustic properties of the target voice. Two Time-Scale Update Rule (TTUR) is also used to facilitate stable training. Experimental results show that the proposed method outperforms previous methods, including the StarGAN-VC network on which it was based.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.28 no.6D
• /
• pp.809-817
• /
• 2008

This paper deals with the development of ViscoElastic Continuum Damage Finite Element Program (VECD-FEP++) and its verification with the results from both field and laboratory accelerated pavement tests. Damage characteristics of asphalt concrete mixture have been defined by Schapery's work potential theory, and uniaxial constant crosshead rate tests were carried out to be used for damage model implementation. VECD-FEP++ predictions were compared with strain responses (longitudinal and transverse strains) under moving wheel loads running at different constant speeds. To this end, an asphalt pavement section (A5) of Korea Expressway Corporation Test Road (KECTR) instrumented with strain gauges were loaded with a dump truck. Also, a series of accelerated pavement fatigue tests have been conducted at pavement sections surfaced with four asphalt concrete mixtures (Dense-graded, SBS, Terpolymer, CR-TB). Planar strain responses were in good agreement with field measurements at base layers, whereas strains at both surface and intermediate layers were found different from simulation results due to the complexity of tire-road contact pressures. Finally, fatigue characteristics of four asphalt mixtures were reasonably described with VECD-FEP++.