• 소성∙가공
• /
• 제31권4호
• /
• pp.229-239
• /
• 2022

Convolution Neural Network(CNN) is a class of deep learning algorithms and can be used for image analysis. In particular, it has excellent performance in finding the pattern of images. Therefore, CNN is commonly applied for recognizing, learning and classifying images. In this study, the surface defect classification performance of Al 6061 extruded material using CNN-based algorithms were compared and evaluated. First, the data collection criteria were suggested and a total of 2,024 datasets were prepared. And they were randomly classified into 1,417 learning data and 607 evaluation data. After that, the size and quality of the training data set were improved using data augmentation techniques to increase the performance of deep learning. The CNN-based algorithms used in this study were VGGNet-16, VGGNet-19, ResNet-50 and DenseNet-121. The evaluation of the defect classification performance was made by comparing the accuracy, loss, and learning speed using verification data. The DenseNet-121 algorithm showed better performance than other algorithms with an accuracy of 99.13% and a loss value of 0.037. This was due to the structural characteristics of the DenseNet model, and the information loss was reduced by acquiring information from all previous layers for image identification in this algorithm. Based on the above results, the possibility of machine vision application of CNN-based model for the surface defect classification of Al extruded materials was also discussed.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2006년 학술대회 논문집 정보 및 제어부문
• /
• pp.120-122
• /
• 2006

The paper concerns the hybrid optimization of fuzzy inference systems that is based on Hierarchical Fair Competition-based Parallel Genetic Algorithms (HFCGA) and information data granulation. The granulation is realized with the aid of the Hard C-means clustering and HFCGA is a kind of multi-populations of Parallel Genetic Algorithms (PGA), and it is used for structure optimization and parameter identification of fuzzy model. It concerns the fuzzy model-related parameters such as the number of input variables to be used, a collection of specific subset of input variables, the number of membership functions, the order of polynomial, and the apexes of the membership function. In the hybrid optimization process, two general optimization mechanisms are explored. Thestructural optimization is realized via HFCGA and HCM method whereas in case of the parametric optimization we proceed with a standard least square method as well as HFCGA method as well. A comparative analysis demonstrates that the proposed algorithm is superior to the conventional methods.

• Nuclear Engineering and Technology
• /
• 제54권1호
• /
• pp.310-317
• /
• 2022

The mass attenuation coefficient is the primary physical parameter to model narrow beam gamma-ray attenuation. A new machine learning based approach is proposed to model gamma-ray shielding behavior of composites alternative to theoretical calculations. Two fuzzy logic algorithms and a neural network algorithm were trained and tested with different mixture ratios of vanadium slag/epoxy resin/antimony in the 0.05 MeV-2 MeV energy range. Two of the algorithms showed excellent agreement with testing data after optimizing adjustable parameters, with root mean squared error (RMSE) values down to 0.0001. Those results are remarkable because mass attenuation coefficients are often presented with four significant figures. Different training data sizes were tried to determine the least number of data points required to train sufficient models. Data set size more than 1000 is seen to be required to model in above 0.05 MeV energy. Below this energy, more data points with finer energy resolution might be required. Neuro-fuzzy models were three times faster to train than neural network models, while neural network models depicted low RMSE. Fuzzy logic algorithms are overlooked in complex function approximation, yet grid partitioned fuzzy algorithms showed excellent calculation efficiency and good convergence in predicting mass attenuation coefficient.

• 한국측량학회지
• /
• 제30권4호
• /
• pp.363-368
• /
• 2012

선형 단순화 알고리듬의 결과물은 단순화 알고리듬의 선택, 임계값의 선택, 대상 객체의 선택 등에 의해 달라진다. 본 연구에서는 Sleeve-fitting, Visvalingam-Whyatt, Bend-simplify 알고리듬의 임계값을 5단계로 조절하여 건물, 하천, 도로 객체에 적용하고, 단순화 결과물에 대해 거리오차, 면적오차, 각오차를 측정하였다. 그리고 오차값들의 추세선을 통해 정성적인 분석을 실시하였다. 실험 결과 선형 단순화 알고리듬에서 임계값의 조절에 따른 오차값의 변화를 알 수 있었고, 선형 단순화 알고리듬들의 임계값의 변화에 따른 특징을 파악할 수 있었다.

• 한국환경과학회지
• /
• 제18권4호
• /
• pp.355-367
• /
• 2009

Comparison study on the land surface temperatures, which are calculated from four different algorithms for MODIS data, was carried out and the characteristics of each algorithm on land surface temperature estimation were also analysed in this study. Algorithms, which are well used for various satellite data analysis, in the comparisons are proposed by Price, Becker and Li, Ulivieri et al., and Wan. Verification of estimated land surface temperature from each algorithm is also performed using observation based regression data. The coefficient of determination ($R^2$) for daytime land surface temperature estimated from Wan's algorithm is higher than that of another algorithms at all seasons and the value of $R^2$ reach on 0.92 at spring. Although $R^2$ for Ulivieri's algorithm is slightly lower than that for Wan's algorithm, the variation pattern of land surface temperature for two algorithms are similar. However, the difference of estimated values among four algorithms become small at the region of high land surface temperature.

• 대한안전경영과학회지
• /
• 제23권4호
• /
• pp.93-104
• /
• 2021

In this paper, we propose a method for diagnosing overload and working load of collaborative robots through performance analysis of machine learning algorithms. To this end, an experiment was conducted to perform pick & place operation while changing the payload weight of a cooperative robot with a payload capacity of 10 kg. In this experiment, motor torque, position, and speed data generated from the robot controller were collected, and as a result of t-test and f-test, different characteristics were found for each weight based on a payload of 10 kg. In addition, to predict overload and working load from the collected data, machine learning algorithms such as Neural Network, Decision Tree, Random Forest, and Gradient Boosting models were used for experiments. As a result of the experiment, the neural network with more than 99.6% of explanatory power showed the best performance in prediction and classification. The practical contribution of the proposed study is that it suggests a method to collect data required for analysis from the robot without attaching additional sensors to the collaborative robot and the usefulness of a machine learning algorithm for diagnosing robot overload and working load.

• International Journal of Advanced Culture Technology
• /
• 제12권1호
• /
• pp.242-248
• /
• 2024

Contemporary societal and environmental transformations coincide with the emergence of novel mental health challenges. anxiety disorder, a chronic and highly debilitating illness, presents with diverse clinical manifestations. Epidemiological investigations indicate a global prevalence of 5%, with an additional 10% exhibiting subclinical symptoms. Notably, 9% of adolescents demonstrate clinical features. Untreated, anxiety disorder exerts profound detrimental effects on individuals, families, and the broader community. Therefore, it is very meaningful to predict anxiety disorder through machine learning algorithm analysis model. The main research content of this paper is the analysis of the prediction model of anxiety disorder by machine learning algorithms. The research purpose of machine learning algorithms is to use computers to simulate human learning activities. It is a method to locate existing knowledge, acquire new knowledge, continuously improve performance, and achieve self-improvement by learning computers. This article analyzes the relevant theories and characteristics of machine learning algorithms and integrates them into anxiety disorder prediction analysis. The final results of the study show that the AUC of the artificial neural network model is the largest, reaching 0.8255, indicating that it is better than the other two models in prediction accuracy. In terms of running time, the time of the three models is less than 1 second, which is within the acceptable range.

• Smart Structures and Systems
• /
• 제17권6호
• /
• pp.935-956
• /
• 2016

With the help of advanced image acquisition and processing technology, the vision-based measurement methods have been broadly applied to implement the structural monitoring and condition identification of civil engineering structures. Many noncontact approaches enabled by different digital image processing algorithms are developed to overcome the problems in conventional structural dynamic displacement measurement. This paper presents three kinds of image processing algorithms for structural dynamic displacement measurement, i.e., the grayscale pattern matching (GPM) algorithm, the color pattern matching (CPM) algorithm, and the mean shift tracking (MST) algorithm. A vision-based system programmed with the three image processing algorithms is developed for multi-point structural dynamic displacement measurement. The dynamic displacement time histories of multiple vision points are simultaneously measured by the vision-based system and the magnetostrictive displacement sensor (MDS) during the laboratory shaking table tests of a three-story steel frame model. The comparative analysis results indicate that the developed vision-based system exhibits excellent performance in structural dynamic displacement measurement by use of the three different image processing algorithms. The field application experiments are also carried out on an arch bridge for the measurement of displacement influence lines during the loading tests to validate the effectiveness of the vision-based system.

• 컴퓨터교육학회논문지
• /
• 제14권5호
• /
• pp.1-8
• /
• 2011

본 연구에서는 전문계 고등학교 학생들의 알고리즘 시간효율성에 관한 개념 이해 수준을 확인하고 빈번하게 발생하는 오개념 유형들을 확인하고자 하였다. 선행연구에서 확인된 세 가지 유형의 오개념들을 토대로 9개의 오개념 측정 문항을 개발하였으며, 프로그래밍 관련 과목을 이수한 51명의 전문계 고등학교 학생들을 대상으로 조사를 실시하였다. 연구결과, 알고리즘 시간효율성에 관한 개념 이해도는 매우 낮은 수준으로 나타났으며, 견고하고 반복적으로 발생되는 여러 유형의 오개념들을 확인하였다. 마지막으로, 확인된 오개념들을 과학적 개념으로 변화시키기 위한 교육적 전략인 시뮬레이션 및 알고리즘 효율성 평가 틀의 활용 방안과 함께 교육과정 개선의 필요성을 제안하였다.

• 상하수도학회지
• /
• 제29권3호
• /
• pp.415-425
• /
• 2015

Aging water pipe networks hinder efficient management of important water service indices such as revenue water and leakage ratio due to pipe breakage and malfunctioning of pipe appurtenance. In order to control leakage in water pipe networks, various methods such as the minimum night flow analysis and sound waves method have been used. However, the accuracy and efficiency of detecting water leak by these methods need to be improved due to the increase of water consumption at night. In this study the Principal Component Analysis (PCA) technique was applied to the night water flow data of 426 days collected from a water distribution system in the interval of one hour. Based on the PCA technique, computational algorithms were developed to narrow the time windows for efficient execution of leak detection job. The algorithms were programmed on computer using the MATLAB. The presented techniques are expected to contribute to the efficient management of water pipe networks by providing more effective time windows for the detection of the anomaly of pipe network such as leak or abnormal demand.