• ETRI Journal
• /
• v.45 no.2
• /
• pp.318-328
• /
• 2023

Recently, embedded systems, such as mobile platforms, have multiple processing units that can operate in parallel, such as centralized processing units (CPUs) and neural processing units (NPUs). We can use deep-learning compilers to generate machine code optimized for these embedded systems from a deep neural network (DNN). However, the deep-learning compilers proposed so far generate codes that sequentially execute DNN operators on a single processing unit or parallel codes for graphic processing units (GPUs). In this study, we propose PartitionTuner, an operator scheduler for deep-learning compilers that supports multiple heterogeneous PUs including CPUs and NPUs. PartitionTuner can generate an operator-scheduling plan that uses all available PUs simultaneously to minimize overall DNN inference time. Operator scheduling is based on the analysis of DNN architecture and the performance profiles of individual and group operators measured on heterogeneous processing units. By the experiments for seven DNNs, PartitionTuner generates scheduling plans that perform 5.03% better than a static type-based operator-scheduling technique for SqueezeNet. In addition, PartitionTuner outperforms recent profiling-based operator-scheduling techniques for ResNet50, ResNet18, and SqueezeNet by 7.18%, 5.36%, and 2.73%, respectively.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.3E
• /
• pp.105-111
• /
• 2002

The parabolic equation technique provides an excellent model to describe the wave phenomena when there exists a predominant direction of propagation. The model handles the square root wave number operator in paraxial direction. Realization of the pseudo-differential square root operator is the essential part of the parabolic equation method for its numerical accuracy. The wide-angled approximation of the operator is made based on the Pade series expansion, where the branch line rotation scheme can be combined with the original Pade approximation to stabilize its computational performance for complex modes. The Galerkin integration has been employed to discretize the depth-dependent operator. The benchmark tests involving the half-infinite space, the range independent and dependent environment will validate the implemented numerical model.

• Journal of the Korean Society of Systems Engineering
• /
• v.16 no.2
• /
• pp.16-26
• /
• 2020

Improving operator performance in complex and time-critical situations is critical to maintain plant safety and operability. These situations require quick detection, diagnosis, and mitigation actions to recover from the root cause of failure. One of the key challenges for operators in nuclear power plants is information management and following the control procedures and instructions. Nowadays Digital Twin technology can be used for analyzing and fast detection of failures and transient situations with the recommender system to provide the operator or maintenance engineer with recommended action to be carried out. Systems engineering approach (SE) is used in developing a digital twin for the CEDM system to support operator actions when there is a misalignment in the control element assembly group. Systems engineering is introduced for identifying the requirements, operational concept, and associated verification and validation steps required in the development process. The system developed by using a machine learning algorithm with a text mining technique to extract the required actions from limiting conditions for operations (LCO) or procedures that represent certain tasks.

• Nuclear Engineering and Technology
• /
• v.44 no.4
• /
• pp.415-420
• /
• 2012

It is important to understand the amount of time required to execute an emergency procedural task in a high-stress situation for managing human performance under emergencies in a nuclear power plant. However, the time to execute an emergency procedural task is highly dependent upon expert judgment due to the lack of actual data. This paper proposes an analytical method to estimate the operator's performance time (OPT) of a procedural task, which is based on a measure of the task complexity (TACOM). The proposed method for estimating an OPT is an equation that uses the TACOM as a variable, and the OPT of a procedural task can be calculated if its relevant TACOM score is available. The validity of the proposed equation is demonstrated by comparing the estimated OPTs with the observed OPTs for emergency procedural tasks in a steam generator tube rupture scenario.

• Journal of Institute of Control, Robotics and Systems
• /
• v.11 no.8
• /
• pp.709-716
• /
• 2005

As a way to build more efficient and intelligent container cranes for todays hub ports, communication networks are used to interconnect numerous sensors, actuators, controllers, and operator switches and consoles that are spatially distributed over a crane. Various signals such as sensor values and operator's commands are digitized and broadcast on the network instead of using separate wiring cables. This not only makes the design and manufacturing of a crane more efficient, but also easier implementation of intelligent control algorithms. This paper presents the performance evaluation of CAN(Controller Area Network), TTP(Time Triggered Protocol) and Byteflight that can be used for cranes. Through discrete event simulation, several important quantitative performance factors such as the probability of a transmission failure, average system delay (data latency) and maximum system delay have been evaluated.

• The Journal of the Korea Contents Association
• /
• v.20 no.12
• /
• pp.287-300
• /
• 2020

The present study applied the IPA(Importance-Performance Analysis) to derive several practical alternatives for the improvement of Chuncheon Silver Orchestra program using IPA(importance-performance analysis). Most importantly, this work performed a comparative study using sample data collected from two groups of silver member(N=53) and program operator(N=32). A careful literature review leaded to the development of six types of evaluation sectors, including 'connection with the local community', 'budget operation', 'program operation', 'manpower operation', 'public relations strategy' as well as 'environment for program operation'. This work further devised twenty-two evaluation items. The results indicated that silver member group rated 'manpower operation' the highest in their both perceived importance and performance. Program operator group rated 'budget operation' and 'manpower operation' rated the highest in their perceived importance and performance, respectively. The results form IPA matrix analysis provides an important implication that much efforts and interests should be sustainably given from the perspective of the silver member group and the program operator group coupled with four types of IPA quadrant, accordingly.

• Journal of Korean Institute of Industrial Engineers
• /
• v.26 no.1
• /
• pp.54-65
• /
• 2000

The objective of this study is to evaluate the operator performance relating to hand switches with two or three buttons in the main control rooms of nuclear power plants. Based on the comparative analysis of the nuclear power plants, two different subjective workload-rating scales were used to evaluate the performance of 48 operators: the Overall Workload(OW) and National Aeronautics and Space Administration Task Load Index (NASA TLX). The survey questions consisting of the eight-items were asked to evaluate the operating experiences for the two different switch types. The OW scales ratings were applied to measure the workload of the switch-related tasks. The ratings revealed that signal detection tasks caused less workload in the three-buttoned-switch operators than the other switch group. However, in the switch operation tasks, the switch types did not show statistically significant effects on workload level. The NASA TLX scale ratings were performed based on detailed task scenarios that assumed the accident of small break loss of coolant, what we call, the small LOCH. The NASA TLX was administered to three different task groups: the reactor, the turbine, and the electric operator groups. Based on the NASA TLX, the two-buttoned switch groups showed higher workload than those with the three-buttoned switches. However, a statistically significant difference was found only in the reactor operator groups. When the current switch type was assumed to be changed for the other type, all of the three-buttoned switch groups were predicted to have higher workload than the other switch groups, respectively. The implications of these findings were discussed.

• IE interfaces
• /
• v.22 no.2
• /
• pp.104-115
• /
• 2009

The objective of the study is to evaluate the effectiveness of Crew Resource Management (CRM) training program for operators in the Main Control Room (MCR) simulator of APR-1400 Nuclear Power Plant. The experiments were conducted for two different crews of operators performing six different emergency operating scenarios during four-week period. Each crew consisted of the five operators: senior reactor operator, safety technical advisor, reactor operator, turbine operator, and electric operator. All crews (Crew A and B) participated in the training program for the technical knowledge and skills which were required to operate the simulator of the MCR during the first week. To verify the effectiveness of the CRM training program; however, only Crew A was selected to attend the CRM training after the technical knowledge and skills training. The results of the experiments showed that the CRM training program improved the individual attitudes of Crew A significantly. Team skills of Crew A were found to be significantly better than those of Crew B. The CRM training did not have positive effects on enhancing the individual performance of Crew A; however, as compared to that of Crew B. Implication of these findings was discussed further in detail.

• Proceedings of the IEEK Conference
• /
• 2005.11a
• /
• pp.979-982
• /
• 2005

This paper is to be pre-processing that decides the text recognizability and quality contained in natural image. Differentiated with the existing studies, In this paper, it suggests the application of partially unified color models, Coiflet Wavelet and text extraction algorithm that uses the closed curve edge features of LoG (laplacian of gaussian)operator. The text image included in natural image such as signboard has the same hue, saturation and value, and there is a certain thickness as for their feature. Each color element is restructured into closed area by LoG operator, the 2nd differential operator. The text area is contracted by Hough Transform, logical AND-OR operator of each color model and Minimum-Distance classifier. This paper targets natural image into which text area is added regardless of the size and resolution of the image, and it is confirmed to have more excellent performance than other algorithms with many restrictions.

• Nuclear Engineering and Technology
• /
• v.52 no.5
• /
• pp.984-994
• /
• 2020

It is reported that about 20% of accidents at nuclear power plants in Korea and abroad are caused by human error. One of the main factors contributing to human error is fatigue, so it is necessary to prevent human errors that may occur when the task is performed in an improper state by grasping the status of the operator in advance. In this study, we propose a method of evaluating operator's fitness-for-duty (FFD) using various parameters including eye movement data, subjective fatigue ratings, and operator's performance. Parameters for evaluating FFD were selected through a literature survey. We performed experiments that test subjects who felt various levels of fatigue monitor information of indicators and diagnose a system malfunction. In order to find meaningful characteristics in measured data consisting of various parameters, hierarchical clustering analysis, an unsupervised machine-learning technique, is used. The characteristics of each cluster were analyzed; fitness-for-duty of each cluster was evaluated. The appropriateness of the number of clusters obtained through clustering analysis was evaluated using both the Elbow and Silhouette methods. Finally, it was statistically shown that the suggested methodology for evaluating FFD does not generate additional fatigue in subjects. Relevance to industry: The methodology for evaluating an operator's fitness for duty in advance is proposed, and it can prevent human errors that might be caused by inappropriate condition in nuclear industries.