• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.81-84
• /
• 2022

제조 산업에서의 이상치 검출은 생산품의 품질과 운영비용을 절감하기 위한 중요한 요소로 최근 딥러닝을 사용하여 자동화되고 있다. 이상치 검출을 위한 딥러닝 기법에는 CNN이 있으며, CNN을 계층적으로 구성할 경우 단일 CNN 모델에 비해 상대적으로 성능의 향상을 보일 수 있다는 것이 많은 선행 연구에서 나타났다. 이에 MVTec-AD 데이터셋을 이용하여 계층 CNN이 다중 클래스 이상치 판별 문제에 대해 효과적인지를 탐구하고자 하였다. 실험 결과 단일 CNN의 정확도는 0.7715, 계층 CNN의 정확도는 0.7838로 다중 클래스 이상치 판별 문제에 있어 계층 CNN 방식 접근이 다중 클래스 이상치 탐지 문제에서 알고리즘의 성능을 향상할 수 있음을 확인할 수 있었다. 계층 CNN은 모델과 파라미터의 개수와 리소스의 사용이 단일 CNN에 비하여 기하급수적으로 증가한다는 단점이 존재한다. 이에 계층 CNN의 장점을 유지하며 사용 리소스를 절약하고자 하였고 K-means, GMM, 계층적 클러스터링 알고리즘을 통해 제작한 새로운 클래스를 이용해 계층 CNN을 구성하여 각각 정확도 0.7930, 0.7891, 0.7936의 결과를 얻을 수 있었다. 이를 통해 Clustering 알고리즘을 사용하여 적절히 물체를 분류할 경우 물체에 따른 개별 상태 판단 모델을 제작하는 것과 비슷하거나 더 좋은 성능을 내며 리소스 사용을 줄일 수 있음을 확인할 수 있었다.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.556-558
• /
• 2022

This paper proposes a CanSat system with a vehicle tracking function based on Jetson Nano, a high-performance small computer capable of operating artificial intelligence algorithms. The CanSat system consists of a CanSat and a ground station. The CanSat falls in the atmosphere and transmits the data obtained through the installed sensors to the ground station using wireless communication. The existing CanSat is limited to the mission of simply transmitting the collected information to the ground station, and there is a limit to efficiently performing the mission due to the limited fall time and bandwidth limitation of wireless communication. The Jetson Nano based CanSat proposed in this paper uses a pre-trained neural network model to detect the location of a vehicle in each image taken from the air in real time, and then uses a 2-axis motor to move the camera to track the vehicle.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2022.10a
• /
• pp.570-573
• /
• 2022

This paper presents the identification of void diameters for a cast-resin transformer using an artificial neural network (ANN) model. A PD signal was measured by the Rogowski coil sensor which has the planar and thin structures fabricated on a printed circuit board (PCB), and the PD electrode system was fabricated to simulate a PD defect by a void. In addition, void samples with different diameters were fabricated by injecting air in a cylindrical aluminum frame using a syringe during the epoxy curing process. To identify the diameter of void defects, PD characteristics such as the discharge magnitude, pulse count, and phase angle were extracted and back propagation algorithm (BPA) was designed using virtual instrument (VI) based on the Labview program. From the experimental results, the BPA algorithm proposed in this paper has over 90% accurate rate to identify the diameter of void defects and is expected to use reference data of maintenance and replacement of insulation for cast-resin transformers in the on-site PD measurement.

• Earthquakes and Structures
• /
• v.23 no.2
• /
• pp.129-138
• /
• 2022

One of the objectives to prevent building pounding between two adjacentstructures is to considerseparation distance or decrease relative displacement during seismic excitation. Although the majority of building codes around the world have basically suggested some equations or approximately recommended various distances between structuresto avoid pounding hazard, but a lot of reportsin zone of pounding have obviously shown thatsafety situation or economic consideration are not always provided due to the collisions between buildings and the cost of land, respectively. For this purpose, a dynamic MDOF model by having base isolation system is numerically considered and using various earthquake records, relative displacements are mathematically investigated. Different equations to determine the value of damping ratio are collected and the results of evaluations are listed for comparison among them to present a new equation for determination of impact damping ratio. Presented equation is depends significantly on impact velocity before and after impact based on artificial neural network, which the accuracy of them is investigated and also confirmed. In order to select the optimum equation, hysteresisloop of impact between base of building and rubber bumper is considered and compared with the hysteresis loop of each impact, calculated by different equations. Finally, using representative equation, the effect of thickness, number and stiffness of rubber bumpers are numerically investigated. The results of analysis indicate that stiffness and number of bumpers have significantly affected in zone of impact force while the thickness of bumpers have not shown significant influence to calculate impact force during earthquake. For instance, increasing the number of bumpers, gap size between structures and also the value of stiffness is caused to decrease impact force between models. The final evaluation demonstrates that bumpers are able to decrease peak lateral displacement of top story during impact.

• Geomechanics and Engineering
• /
• v.31 no.2
• /
• pp.129-147
• /
• 2022

The properties of soil are naturally highly variable and thus, to ensure proper safety and reliability, we need to test a large number of samples across the length and depth. In pile foundations, conducting field tests are highly expensive and the traditional empirical relations too have been proven to be poor in performance. The study proposes a state-of-art Particle Swarm Optimization (PSO) hybridized Artificial Neural Network (ANN), Extreme Learning Machine (ELM) and Adaptive Neuro Fuzzy Inference System (ANFIS); and comparative analysis of metaheuristic models (ANN-PSO, ELM-PSO, ANFIS-PSO) for prediction of bearing capacity of pile foundation trained and tested on dataset of nearly 300 dynamic pile tests from the literature. A novel ensemble model of three hybrid models is constructed to combine and enhance the predictions of the individual models effectively. The authenticity of the dataset is confirmed using descriptive statistics, correlation matrix and sensitivity analysis. Ram weight and diameter of pile are found to be most influential input parameter. The comparative analysis reveals that ANFIS-PSO is the best performing model in testing phase (R² = 0.85, RMSE = 0.01) while ELM-PSO performs best in training phase (R² = 0.88, RMSE = 0.08); while the ensemble provided overall best performance based on the rank score. The performance of ANN-PSO is least satisfactory compared to the other two models. The findings were confirmed using Taylor diagram, error matrix and uncertainty analysis. Based on the results ELM-PSO and ANFIS-PSO is proposed to be used for the prediction of bearing capacity of piles and ensemble learning method of joining the outputs of individual models should be encouraged. The study possesses the potential to assist geotechnical engineers in the design phase of civil engineering projects.

• International conference on construction engineering and project management
• /
• 2020.12a
• /
• pp.463-481
• /
• 2020

Cutter suction dredgers (CSDs) are widely used in various dredging constructions such as channel excavation, wharf construction, and reef construction. During a CSD construction, the main operation is to control the swing speed of cutter to keep the slurry concentration in a proper range. However, the slurry concentration cannot be monitored in real-time, i.e., there is a "time-lag effect" in the log of slurry concentration, making it difficult for operators to make the optimal decision on controlling. Concerning this issue, a solution scheme that using real-time monitored indicators to predict current slurry concentration is proposed in this research. The characteristics of the CSD monitoring data are first studied, and a set of preprocessing methods are presented. Then we put forward the concept of "index class" to select the important indices. Finally, an ensemble learning algorithm is set up to fit the relationship between the slurry concentration and the indices of the index classes. In the experiment, log data over seven days of a practical dredging construction is collected. For comparison, the Deep Neural Network (DNN), Long Short Time Memory (LSTM), Support Vector Machine (SVM), Random Forest (RF), Gradient Boosting Decision Tree (GBDT), and the Bayesian Ridge algorithm are tried. The results show that our method has the best performance with an R² of 0.886 and a mean square error (MSE) of 5.538. This research provides an effective way for real-time predicting the slurry concentration of CSDs and can help to improve the stationarity and production efficiency of dredging construction.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• 2023.05a
• /
• pp.104-105
• /
• 2023

해양 관측과 위성 원격탐사를 이용하여 시공간적으로 다양하게 변하는 생태 어장 환경 및 선박 관련 자료를 획득할 수 있다. 이번 연구의 주요 목적은 야간 불빛 위성 자료를 이용하여 광범위한 해역에 대한 어선의 위치 분포를 파악하는 딥러닝 기반 모델을 제안하는 것이다. 제안한 모델의 정확성을 평가하기 위해 야간 조업 어선의 위치를 포함하고 있는 AIS(Automatic Identification System) 정보와 상호 비교 평가 하였다. 이를 위해, 먼저 AIS 자료를 획득 및 분석하는 방법을 소개한다. 해양안전종합시스템(General Information Center on Maritime Safety & Security, GICOMS)으로부터 제공받은 AIS 자료는 동적정보와 정적정보로 나뉜다. 동적 정보는 일별 자료로 구분되어있으며, 이 정보에는 해상이동업무식별번호(Maritime Mobile Service Identity, MMSI), 선박의 시간, 위도, 경도, 속력(Speed over Ground, SOG), 실침로(Course over Ground, COG), 선수방향(Heading) 등이 포함되어 있다. 정적정보는 1개의 파일로 구성되어 있으며, 선박명, 선종 코드, IMO Number, 호출부호, 제원(DimA, DimB, DimC, Dim D), 홀수, 추정 톤수 등이 포함되어 있다. 이번 연구에서는 선박의 정보에서 어선의 정보를 추출하여 비교 자료로 사용하였으며, 위성 자료는 구름의 영향이 없는 깨끗한 날짜의 영상 자료를 선별하여 사용하였다. 야간 불빛 위성 자료, 구름 정보 등을 이용하여 야간 조업 어선의 불빛을 감지하는 심층신경망(Deep Neural Network; DNN) 기반 모델을 제안하였다. 본 연구의결과는 야간 어선의 분포를 감시하고 한반도 인근 어장을 보호하는데 기여할 것으로 기대된다.

• Journal of IKEEE
• /
• v.27 no.4
• /
• pp.601-609
• /
• 2023

In this paper, we propose an artificial intelligence-based high-speed fault diagnosis method using an FPGA in the event of a ground fault in a DC system. When applying artificial intelligence algorithms to fault diagnosis, a substantial amount of computation and real-time data processing are required. By employing an FPGA with AI-based high-speed fault diagnosis, the DC breaker can operate more rapidly, thereby reducing the breaking capacity of the DC breaker. therefore, in this paper, an intelligent high-speed diagnosis algorithm was implemented by collecting fault data through fault simulation of a DC system using Matlab/Simulink. Subsequently, the proposed intelligent high-speed fault diagnosis algorithm was applied to the FPGA, and performance verification was conducted.

• Annual Conference on Human and Language Technology
• /
• 2023.10a
• /
• pp.659-663
• /
• 2023

이 논문은 한국어 음성인식 성능을 개선하고자 기존 음성인식 과정을 자모단위 음성인식 모델과 신경망 기반 자모 병합 모델 총 두 단계로 구성하였다. 한국어는 조합어 특성상 음성 인식에 필요한 음절 단위가 약 2900자에 이른다. 이는 학습 데이터셋에 자주 등장하지 않는 음절에 대해서 음성인식 성능을 저하시키고, 학습 비용을 높이는 단점이 있다. 이를 개선하고자 음절 단위의 인식이 아닌 51가지 자모 단위(ㄱ-ㅎ, ㅏ-ㅞ)의 음성인식을 수행한 후 자모 단위 인식 결과를 음절단위의 한글로 병합하는 과정을 수행할 수 있다[1]. 자모단위 인식결과는 초성, 중성, 종성을 고려하면 규칙 기반의 병합이 가능하다. 하지만 음성인식 결과에 잘못인식된 자모가 포함되어 있다면 최종 병합 결과에 오류를 생성하고 만다. 이를 해결하고자 신경망 기반의 자모 병합 모델을 제시한다. 자모 병합 모델은 분리되어 있는 자모단위의 입력을 완성된 한글 문장으로 변환하는 작업을 수행하고, 이 과정에서 음성인식 결과로 잘못인식된 자모에 대해서도 올바른 한글 문장으로 변환하는 오류 수정이 가능하다. 본 연구는 한국어 음성인식 말뭉치 KsponSpeech를 활용하여 실험을 진행하였고, 음성인식 모델로 Wav2Vec2.0 모델을 활용하였다. 기존 규칙 기반의 자모 병합 방법에 비해 제시하는 자모 병합 모델이 상대적 음절단위오류율(Character Error Rate, CER) 17.2% 와 단어단위오류율(Word Error Rate, WER) 13.1% 향상을 확인할 수 있었다.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.16 no.5
• /
• pp.347-353
• /
• 2023

Measuring the growth of fish in fish farms still uses a laborious method. This method requires a lot of labor and causes stress to the fish, which has a negative impact on mortality. To solve this problem, we propose the Fish Growth Regression System (FGRS), a system to automate the growth of fish. FGRS consists of two modules. The first is a module that detects fish based on Yolo v8, and the second consists of a module that predicts the growth of fish using fish image data and a CNN-based neural network model. As a result of the simulation, the average prediction error before learning was 134.2 days, but after learning, the average error decreased to 39.8 days. It is expected that the system proposed in this paper can be used to predict the growing date and use the growth prediction of fish to contribute to automation in fish farms, resulting in a significant reduction in labor and cost savings.