• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.1-25
• /
• 2020

In this paper, we suggest an application system architecture which provides accurate, fast and efficient automatic gasometer reading function. The system captures gasometer image using mobile device camera, transmits the image to a cloud server on top of private LTE network, and analyzes the image to extract character information of device ID and gas usage amount by selective optical character recognition based on deep learning technology. In general, there are many types of character in an image and optical character recognition technology extracts all character information in an image. But some applications need to ignore non-of-interest types of character and only have to focus on some specific types of characters. For an example of the application, automatic gasometer reading system only need to extract device ID and gas usage amount character information from gasometer images to send bill to users. Non-of-interest character strings, such as device type, manufacturer, manufacturing date, specification and etc., are not valuable information to the application. Thus, the application have to analyze point of interest region and specific types of characters to extract valuable information only. We adopted CNN (Convolutional Neural Network) based object detection and CRNN (Convolutional Recurrent Neural Network) technology for selective optical character recognition which only analyze point of interest region for selective character information extraction. We build up 3 neural networks for the application system. The first is a convolutional neural network which detects point of interest region of gas usage amount and device ID information character strings, the second is another convolutional neural network which transforms spatial information of point of interest region to spatial sequential feature vectors, and the third is bi-directional long short term memory network which converts spatial sequential information to character strings using time-series analysis mapping from feature vectors to character strings. In this research, point of interest character strings are device ID and gas usage amount. Device ID consists of 12 arabic character strings and gas usage amount consists of 4 ~ 5 arabic character strings. All system components are implemented in Amazon Web Service Cloud with Intel Zeon E5-2686 v4 CPU and NVidia TESLA V100 GPU. The system architecture adopts master-lave processing structure for efficient and fast parallel processing coping with about 700,000 requests per day. Mobile device captures gasometer image and transmits to master process in AWS cloud. Master process runs on Intel Zeon CPU and pushes reading request from mobile device to an input queue with FIFO (First In First Out) structure. Slave process consists of 3 types of deep neural networks which conduct character recognition process and runs on NVidia GPU module. Slave process is always polling the input queue to get recognition request. If there are some requests from master process in the input queue, slave process converts the image in the input queue to device ID character string, gas usage amount character string and position information of the strings, returns the information to output queue, and switch to idle mode to poll the input queue. Master process gets final information form the output queue and delivers the information to the mobile device. We used total 27,120 gasometer images for training, validation and testing of 3 types of deep neural network. 22,985 images were used for training and validation, 4,135 images were used for testing. We randomly splitted 22,985 images with 8:2 ratio for training and validation respectively for each training epoch. 4,135 test image were categorized into 5 types (Normal, noise, reflex, scale and slant). Normal data is clean image data, noise means image with noise signal, relfex means image with light reflection in gasometer region, scale means images with small object size due to long-distance capturing and slant means images which is not horizontally flat. Final character string recognition accuracies for device ID and gas usage amount of normal data are 0.960 and 0.864 respectively.

• Bioinformatics and Biosystems
• /
• v.1 no.2
• /
• pp.95-98
• /
• 2006

Recent proteomic studies of protein domains require high-throughput and systematic approaches. Since most experiments using protein domains, the modules of protein-protein interactions, require gene cloning, the first experimental step should be retrieving DNA sequences of domain encoding regions from databases. For a large scale proteomic research, however, it is a laborious task to extract a large number of domain sequences manually from several inter-linked databases. We present a new methodology to retrieve DNA sequences of domain encoding regions through automatic database cross-referencing. To extract protein domain encoding regions, it traverses several inter-connected database with validation process. And we applied this method to retrieve all the EGF domain encoding DNA sequences of homo sapiens. This new algorithm was implemented using Python library PAMIE, which enables to cross-reference across distinct databases automatically.

• Smart Structures and Systems
• /
• v.11 no.5
• /
• pp.477-496
• /
• 2013

Due to their cost-effectiveness and ease of installation, wireless smart sensors (WSS) have received considerable recent attention for structural health monitoring of civil infrastructure. Though various wireless smart sensor networks (WSSN) have been successfully implemented for full-scale structural health monitoring (SHM) applications, monitoring of low-level ambient strain still remains a challenging problem for WSS due to A/D converter (ADC) resolution, inherent circuit noise, and the need for automatic operation. In this paper, the design and validation of high-precision strain sensor board for the Imote2 WSS platform and its application to SHM of a cable-stayed bridge are presented. By accurate and automated balancing of the Wheatstone bridge, signal amplification of up to 2507-times can be obtained, while keeping signal mean close to the center of the ADC span, which allows utilization of the full span of the ADC. For better applicability to SHM for real-world structures, temperature compensation and shunt calibration are also implemented. Moreover, the sensor board has been designed to accommodate a friction-type magnet strain sensor, in addition to traditional foil-type strain gages, facilitating fast and easy deployment. The wireless strain sensor board performance is verified through both laboratory-scale tests and deployment on a full-scale cable-stayed bridge.

• Journal of the Korea Institute of Military Science and Technology
• /
• v.18 no.3
• /
• pp.335-342
• /
• 2015

Laboratory facilities for biology are designed as biosafety level 1, biosafety level 2, biosafety level 3, and biosafety level 4. Biosafety level designations are based on a composite of the design features, construction, containment facilities, equipment, practice and operation procedures required for working with agents from the various risk groups. Generally, biosafety level 3 means the facility that is appropriate for the experiments using pathogens which can cause serious diseases by aerosol transmission. The biosafety level assigned for the specific work to be done is driven by professional judgement based on a risk assessment, rather than by automatic assignment according to the particular risk group designation of the pathogenic agents to be used. In this paper, we introduced the biosafety level 3 facility operated in ADD(Agency for defense development). It contains the overview of facility, microbiological experiment, animal experiment, decontamination and waste disposal. Biosafety level 3 laboratory in ADD has served the vital role in the research of biological agents and antidote development.

• Journal of the Korean Data and Information Science Society
• /
• v.9 no.2
• /
• pp.179-188
• /
• 1998

Local polynomial regression estimation is the most popular one among kernel type regression estimator. In local polynomial regression function esimation bandwidth selection is crucial problem like the kernel estimation. When the regression curve has complicated structure variable bandwidth selection will be appropriate. In this paper, we propose a variable bandwidth selection method fully data driven. We will choose the bandwdith by selecting minimising estiamted MSE which is estimated by the pilot bandwidth study via croos-validation method. Monte carlo simulation was conducted in order to show the superiority of proposed bandwidth selection method.

• Nuclear Engineering and Technology
• /
• v.31 no.2
• /
• pp.157-171
• /
• 1999

In this work, an automatic software verification method for Nuclear Power Plant (NPP) protection system is developed. This method utilizes Colored Petri Net (CPN) for system modeling and Prototype Verification System (PVS) for mathematical verification. In order to help flow-through from modeling by CPN to mathematical proof by PVS, an information extractor from CPN models has been developed in this work. In order to convert the extracted information to the PVS specification language, a translator also has been developed. ML that is a higher-order functional language programs the information extractor and translator. This combined method has been applied to a protection system function of Wolsong NPP SDS2(Steam Generator Low Level Trip). As a result of this application, we could prove completeness and consistency of the requirement logically. Through this work, in short, an axiom or lemma based-analysis method for CPN models is newly suggested in order to complement CPN analysis methods and a guideline for the use of formal methods is proposed in order to apply them to NPP Software Verification and Validation.

• Journal of Environmental Science International
• /
• v.13 no.7
• /
• pp.637-643
• /
• 2004

A new method of automatic recording raingauge is developed to measure rainfall with 0.01mm resolution. This use two different signals to measure rainfall more accurately compare than other raingauges. One is weight of the tipping bucket with rainfall amount and the other is pulse from tipping bucket reverse. New method applied 1 mm tipping bucket mechanism and install loadcell under tipping bucket mechanism for measuring rainfall weight. Loadcell measure weight of rainfall until 1 mm with 0.01 mm resolution and more than 1 mm than bucket reverse and pulse signal generate, after that loadcell measure weight again. The validation of new instrument was examined in the room 65 mm/hour rainfall rate total 53 mm range. There is below than 1 % error of absolute rainfall amount and 0.01 mm resolution. The field test of instrument was carried out by comparing its measured values with values recorded by weight type and standard type on June 1 2003 at Terrestrial Environmental Research Center at Tsukuba University in Tsukuba of Japan, when it has recorded total amount of 40.58 mm rainfall by standard raingauge and new raingauge recorded 41.032 mm. Same rainfall intensity pattern observed in field observation with weight type raingauge. Rainfall intensity between weight type and Lee-A type raingauge reached 0.9947 correlation in 3 minute average.

• International Journal of Highway Engineering
• /
• v.14 no.4
• /
• pp.141-148
• /
• 2012

PURPOSES : This study demonstrates the need for the collection of road weather information in order to perform efficient snow removal works during the winter season. Snow removal operations are usually dependent upon weather information obtained from the Automatic Weather Station provided by the Korea Meteorological Administration. However, there are some difference between road weather and weather forecasts in their scope. This is because general weather forecasts are focused on macroscopic standpoints rather than microscopic perspectives. METHODS : In this study, the relationship between snow removal works and historical weather forecasts are properly analyzed to prove the importance of road weather information. We collected both weather data and snow removal works during winter season at "A" regional offices in Gangwon areas. RESULTS : Results showed that the validation of weather forecasts for snow removal works were depended on the height difference between AWS location and its neighboring roadway. CONCLUSIONS : Namely, it appears that road weather information should be collected where AWS location and its neighboring roadway have relatively big difference in their heights.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2002.11b
• /
• pp.422-427
• /
• 2002

This paper focuses on the formulation and validation of an automatic strategy for the selection of the locations and directions of strain gauges to capture at best the modal response of a blade in a series of modes. These locations and directions are selected to render the strain measurements as robust as possible with respect to random mispositioning of the gauges and gauge failures. The approach relies on the evaluation of the signal-to-noise ratios of the gauge measurements from finite element strain data and includes the effects of gauge size. A genetic algorithm is used to find the strain gauge locations-directions that lead to the largest possible value of the smallest modal strain signal-to-noise ratio, in the absence of gauge failure, or of its expected value when gauge failure is possible. A fan blade is used to exemplify the applicability of the proposed methodology and to demonstrate the effects of the essential parameters of the problem, i.e. the mispositioning level, the probability of gauge failure, and the number of gauges.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.16 no.2
• /
• pp.8-17
• /
• 2008

This paper presents the Matlab/Simulink-based software-in-the-loop simulation(SILS) environment which is the co-simulator for temporal and functional simulations of control systems. The temporal behavior of a control system is strongly dependent on the implemented software and hardware such as the real-time operating system, the target CPU, and the communication protocol. The proposed SILS abstracts the system with tasks, task executions, real-time schedulers, and real-time networks close to the implementation. Methods to realize these components in graphical block representations are investigated with Matlab/Simulink, which is most commonly used tool for designing and simulating control algorithms in control engineering. In order to achieve a seamless development from SILS to rapid control prototyping (RCP), the SILS block-set is designed to support automatic code generation without tool changes and block modifications.