• International conference on construction engineering and project management
• /
• 2011.02a
• /
• pp.17-22
• /
• 2011

A considerable number of construction disasters occur on pathways. Safety management is usually performed on construction sites to prevent accidents in activity areas. This means that the safety management level of hazards on pathways is relatively minimized. Many researchers have noted that hazard identification is fundamental to safety management. Thus, algorithms for helping safety managers to identify hazardous areas are developed using automated data collection technology. These algorithms primarily search for potential hazardous areas by comparing workers' location logs based on a real-time location system and optimal routes based on BIM. Potential hazardous areas are filtered by identified hazardous areas and activity areas. After that, safety managers are provided with information about potential hazardous areas and can establish proper safety countermeasures. This can help to improve safety on construction sites.

• ETRI Journal
• /
• v.45 no.5
• /
• pp.862-873
• /
• 2023

Determining whether an autonomous self-driving agent is in the middle of an intersection can be extremely difficult when relying on visual input taken from a single camera. In such a problem setting, a wider range of views is essential, which drives us to use three cameras positioned in the front, left, and right of an agent for better intersection recognition. However, collecting adequate training data with three cameras poses several practical difficulties; hence, we propose using data collected from one camera to train a three-camera model, which would enable us to more easily compile a variety of training data to endow our model with improved generalizability. In this work, we provide three separate fusion methods (feature, early, and late) of combining the information from three cameras. Extensive pedestrian-view intersection classification experiments show that our feature fusion model provides an area under the curve and F1-score of 82.00 and 46.48, respectively, which considerably outperforms contemporary three- and one-camera models.

• Journal of Korea Artificial Intelligence Association
• /
• v.1 no.2
• /
• pp.21-25
• /
• 2023

The purpose of this study was to compare the performance using multiple regression models to predict the energy consumption of steel industry. Specific independent variables were selected in consideration of correlation among various attributes such as CO2 concentration, NSM, Week Status, Day of week, and Load Type, and preprocessing was performed to solve the multicollinearity problem. In data preprocessing, we evaluated linear and nonlinear relationships between each attribute through correlation analysis. In particular, we decided to select variables with high correlation and include appropriate variables in the final model to prevent multicollinearity problems. Among the many regression models learned, Boosted Decision Tree Regression showed the best predictive performance. Ensemble learning in this model was able to effectively learn complex patterns while preventing overfitting by combining multiple decision trees. Consequently, these predictive models are expected to provide important information for improving energy efficiency and management decision-making at steel industry. In the future, we plan to improve the performance of the model by collecting more data and extending variables, and the application of the model considering interactions with external factors will also be considered.

• KIPS Transactions on Software and Data Engineering
• /
• v.8 no.12
• /
• pp.491-498
• /
• 2019

Digital twin is a technology that virtualizes physical objects of the real world on a computer. It is used by collecting sensor data through IoT, and using the collected data to connect physical objects and virtual objects in both directions. It has an advantage of minimizing risk by tuning an operation of virtual model through simulation and responding to varying environment by exploiting experiments in advance. Recently, artificial intelligence and machine learning technologies have been attracting attention, so that tendency to virtualize a behavior of physical objects, observe virtual models, and apply various scenarios is increasing. In particular, recognition of each robot's motion is needed to build digital twin for co-robot which is a heart of industry 4.0 factory automation. Compared with modeling based research for recognizing motion of co-robot, there are few attempts to predict motion based on sensor data. Therefore, in this paper, an experimental environment for collecting current and inertia data in co-robot to detect the motion of the robot is built, and a motion prediction model based on the collected sensor data is proposed. The proposed method classifies the co-robot's motion commands into 9 types based on joint position and uses current and inertial sensor values to predict them by accumulated learning. The data used for accumulating learning is the sensor values that are collected when the co-robot operates with margin in input parameters of the motion commands. Through this, the model is constructed to predict not only the nine movements along the same path but also the movements along the similar path. As a result of learning using SVM, the accuracy, precision, and recall factors of the model were evaluated as 97% on average.

• The Journal of the Korea Contents Association
• /
• v.19 no.6
• /
• pp.10-23
• /
• 2019

Although social big data can provide a multi-faceted perspective on school bullying experiences among children and adolescents, the complexity and variety of unstructured text presents a challenge for systematic collection and analysis of the data. Development of an ontology, which identifies key terms and their intricate relationships, is crucial for extracting key concepts and effectively collecting data. The current study elaborated on the definition of an ontology, carefully described the 7 stage development process, and applied the ontology for collecting and analyzing school bullying social big data. As a result, approximately 2,400 key terms were extracted in top-, middle-, and lower-level categories, concerning domains of participants, causes, types, location, region, and intervention. The study contributes to the literature by explaining the ontology development process and proposing a novel alternative research model that uses social big data in school bullying research. Findings from this ontology study may provide a basis for social big data research. Practical implications of this study lie in not only helping to understand the experience of school bullying participants, but also in offering a macro perspective on school bullying as a social phenomenon.

• Culinary science and hospitality research
• /
• v.24 no.2
• /
• pp.34-43
• /
• 2018

The objective of this study was to acquire a general and text-based awareness and recognition of cruise food hygiene through big data analytics. For the purpose, this study collected data with conducting the keyword "food hygiene, cruise" on the web pages and news on Google, during October 1st, 2015 to October 1st, 2017 (two years). The data collection was processed by SCTM which is a data collecting and processing program and eventually, 899 kb, approximately 20,000 words were collected. For the data analysis, UCINET 6.0 packaged with visualization tool-Netdraw was utilized. As a result of the data analysis, the words such as jobs, news, showed the high frequency while the results of centrality (Freeman's degree centrality and Eigenvector centrality) and proximity indicated the distinct rank with the frequency. Meanwhile, as for the result of CONCOR analysis, 4 segmentations were created as "food hygiene group", "person group", "location related group" and "brand group". The diagnosis of this study for the food hygiene in cruise industry through big data is expected to provide instrumental implications both for academia research and empirical application.

• Smart Media Journal
• /
• v.4 no.3
• /
• pp.56-61
• /
• 2015

Although many successful cases regarding big data have been reported, building information systems of big data is still difficult. From the perspective of technology the builders need to understand the whole process of systems development ranging from collecting, storing, processing, and analyzing data to presenting and using information. Whereas, from the perspective of business, the builders need to understand the values of the proposed big data project and explain to top managers who have to make a decision of the risky investment. This study proposes a framework of 5W 1H that can help the builder understand things related to the development of big data information systems. In addition, big data cases from the real world have been illustrated by applying to the framework. It is expected to help builders understand and manage big data projects and lead managers to make better decisions of the investment to the development of information systems.

• Journal of Convergence for Information Technology
• /
• v.9 no.1
• /
• pp.136-141
• /
• 2019

In this paper, we describe the design of a small size data acquisition system with STM32 processor based on Cortex-M4. The system is used for the sensor devices to collect raw data on production lines at factory and send them to the server computer in real time. Also the system is designed to easily acquisite various kinds of data collected from various sensors with the digital signal input unit, the analog signal input unit, the digital signal output unit and the analog signal output unit This small data acquisition system will contribute to the improvement of the quality of precision products in the industrial field by collecting various data in real time and transmitting data at high speed.

• Journal of the Society of Naval Architects of Korea
• /
• v.56 no.2
• /
• pp.143-151
• /
• 2019

This study describes a method for analyzing log data generated in a process using process mining techniques. A system for collecting and analyzing a large amount of log data generated in the process of manufacturing an offshore plant piping material was constructed. The analyzed data was visualized through various methods. Through the analysis of the process model, it was evaluated whether the process performance was correctly input. Through the pattern analysis of the log data, it is possible to check beforehand whether the problem process occurred. In addition, we analyzed the process performance data of partner companies and identified the load of their processes. These data can be used as reference data for pipe production allocation. Real-time decision-making is required to cope with the various variances that arise in offshore plant production. To do this, we have built a system that can analyze the log data of real - time system and make decisions.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.17 no.3
• /
• pp.129-138
• /
• 2022

Collecting and labeling sufficient training data, which is essential to deep learning-based visual inspection, is difficult for manufacturers to perform because it is very expensive. This paper presents a steel plate surface defect detection system with industrial-grade detection performance by training a small amount of steel plate surface images consisting of labeled and non-labeled data. To overcome the problem of lack of training data, we propose two data augmentation techniques: program-based augmentation, which generates defect images in a geometric way, and generative model-based augmentation, which learns the distribution of labeled data. We also propose a 4-step semi-supervised learning using pseudo labels and consistency training with fixed-size augmentation in order to utilize unlabeled data for training. The proposed technique obtained about 99% defect detection performance for four defect types by using 100 real images including labeled and unlabeled data.