• Smart Structures and Systems
• /
• v.23 no.4
• /
• pp.393-403
• /
• 2019

Applying more features gives us better accuracy in modeling; however, increasing the inputs causes the curse of dimensions. In this paper, a new structure has been proposed for fault detecting and identifying (FDI) of high-dimensional systems. This structure consist of two structure. The first part includes Auto-Encoders (AE) as Deep Neural Networks (DNNs) to produce feature engineering process and summarize the features. The second part consists of the Local Model Networks (LMNs) with LOcally LInear MOdel Tree (LOLIMOT) algorithm to model outputs (multiple models). The fault detection is based on these multiple models. Hence the residuals generated by comparing the system output and multiple models have been used to alarm the faults. To show the effectiveness of the proposed structure, it is tested on single-shaft industrial gas turbine prototype model. Finally, a brief comparison between the simulated results and several related works is presented and the well performance of the proposed structure has been illustrated.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.28 no.11
• /
• pp.1741-1751
• /
• 2004

It is commonly requested that the steam generator tubes wall-thinned in excess of 40% should be plugged. However, the plugging criterion is known to be too conservative for some locations and types of defects and its application is limited to a single crack in spite of the fact that the occurrence of multiple through-wall cracks is more common in general. The objective of this research is to propose the optimum failure prediction models for two adjacent through-wall cracks in steam generator tubes. The conservatism of the present plugging criteria was reviewed using the existing failure prediction models for a single crack, and six new failure prediction models for multiple through-wall cracks have been introduced. Then, in order to determine the optimum ones among these new local or global failure prediction models, a series of plastic collapse tests and corresponding finite element analyses for two adjacent through-wall cracks in thin plate were carried out. Thereby, the reaction force model, plastic zone contact model and COD (Crack-Opening Displacement) base model were selected as the optimum ones for assessment of steam generator tubes with multiple through-wall cracks. The selected optimum failure prediction models, finally, were used to estimate the coalescence pressure of two adjacent through-wall cracks in steam generator tubes.

• Journal of Korea Water Resources Association
• /
• v.55 no.3
• /
• pp.229-243
• /
• 2022

The effects of monthly meteorological data measured at 11 stations in South Korea on pan coefficient were analyzed to develop the four types of multiple linear regression models for estimating pan coefficients. To evaluate the applicability of developed models, the models were compared with six previous models. Pan coefficients were most affected by air temperature for January, February, March, July, November and December, and by solar radiation for other months. On the whole, for 12 months of the year, the effects of wind speed and relative humidity on pan coefficient were less significant, compared with those of air temperature and solar radiation. For all meteorological stations and months, the model developed by applying 5 independent variables (wind speed, relative humidity, air temperature, ratio of sunshine duration and daylight duration, and solar radiation) for each station was the most effective for evaporation estimation. The model validation results indicate that the multiple linear regression models can be applied to some particular stations and months.

• Computers and Concrete
• /
• v.20 no.1
• /
• pp.111-118
• /
• 2017

The present study established prediction models based on multiple nonlinear regressions (MNLRs) and backpropagation neural networks (BPNs) for the expansion of cement mortar caused by oxidization slag that was used as a replacement of the aggregate. The data used for the models were obtained from actual laboratory tests on specimens that were produced with water/cement ratios of 0.485 or 1.5, within which 0%, 10%, 20%, 30%, 40%, or 50% of the cement had been replaced by oxidization slag from electric-arc furnaces; the samples underwent high-temperature curing at either $80^{\circ}C$ or $100^{\circ}C$ for 1-4 days. The varied mixing ratios, curing conditions, and water/cement ratios were all used as input parameters for the expansion prediction models, which were subsequently evaluated based on their performance levels. Models of both the MNLR and BPN groups exhibited $R^2$ values greater than 0.8, indicating the effectiveness of both models. However, the BPN models were found to be the most accurate models.

• Journal of information and communication convergence engineering
• /
• v.21 no.4
• /
• pp.346-350
• /
• 2023

Trajectory prediction is an essential element for driving autonomous vehicles, and various trajectory prediction models have emerged with the development of deep learning technology. Convolutional neural network (CNN) is the most commonly used neural network architecture for extracting the features of visual images, and the latest models exhibit high performances. This study was conducted to identify an efficient CNN backbone model among the components of deep learning models for trajectory prediction. We changed the existing CNN backbone network of multiple-trajectory prediction models used as feature extractors to various state-of-the-art CNN models. The experiment was conducted using nuScenes, which is a dataset used for the development of autonomous vehicles. The results of each model were compared using frequently used evaluation metrics for trajectory prediction. Analyzing the impact of the backbone can improve the performance of the trajectory prediction task. Investigating the influence of the backbone on multiple deep learning models can be a future challenge.

• The Bulletin of The Korean Astronomical Society
• /
• v.43 no.1
• /
• pp.46.1-46.1
• /
• 2018

In order to investigate the origin of multiple stellar populations in the halo and bulge of the Milky Way, we have constructed chemical evolution models for the low-mass proto-Galactic subsystems such as globular clusters. Unlike previous studies, we assume that supernova blast waves undergo blowout without expelling the pre-enriched gas, while relatively slow winds of massive stars, together with the winds and ejecta from low and intermediate mass asymptotic-giant-branch stars, are all locally retained in these less massive systems. We find that the observed Na-O anti-correlations in metal-poor GCs can be reproduced when multiple episodes of starbursts are allowed to continue in these subsystems. A specific form of star formation history with decreasing time intervals between the stellar generations, however, is required to obtain this result, which is in good agreement with the parameters obtained from our stellar evolution models for the horizontal-branch. The "mass budget problem" is also much alleviated by our models without ad-hoc assumptions on star formation efficiency and initial mass function. We also applied these models to investigate the origin of super helium-rich red clump stars in the metal-rich bulge as recently suggested by Lee et al. (2015). We find that chemical enrichments by the winds of massive stars can naturally reproduce the required helium enhancement (dY/dZ = 6) for the second-generation stars. Disruption of proto-globular clusters in a hierarchical merging paradigm would have provided helium enhanced stars to the bulge field.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.37 no.1
• /
• pp.1-18
• /
• 2012

The present study provides some extensions over a recent work in Won (2011) which investigates properties of the static newsvendor problem under a schedule involving progressive multiple discounts under the assumption that demand is given exogenously. Khouja (1995, 1996) formulated the extended versions over the classical newsvendor model with various discount policies including all-units discount and/or multiple discounts and found that the extended newsvendor models with discount schedules yield higher optimal expected profits than the classical newsvendor model with no-discounts. In this study, we establish a robust conjecture as a stronger statement than Khouja's findings with regard to the general relationship among the expected profits of newsvendor models in the sense that the conjecture holds for every order quantity as well as the optimal order quantity. The conjecture encourages the newsvendor facing quantity discounts to safely implement her own discounts policy to customer or accept quantity discounts offered by the supplier even if the optimal order quantity cannot be ordered due to additional restrictions such as budget or warehouse capacity constraints because the newsvendor models with quantity discounts always yield higher expected profit than the classic newsvendor model without quantity discounts regardless of the order quantity. Results from wide experiments with various probability distributions of demand strongly support our conjecture.

• Asia pacific journal of information systems
• /
• v.10 no.2
• /
• pp.51-69
• /
• 2000

Recently, as business problems become more complicated and require more precise quantitative results, large-scale model management systems are increasingly in demand for supporting the decision-making activities. In addition, as distributed computing over networks gains popularity, departmental computing systems are gradually adopted in an organization to facilitate collaboration of geographically dispersed multiple departments. In departmental collaborative model management systems, multiple departments share common models but approach them with different user-views depending on their departmental needs. Moreover, the shared models become evolved as their structures and the corresponding data sets change due to the dynamic nature of the operating environment and the inherent uncertainty associated with the problems. In such capacity, providing the multiple departmental users with synchronized and consistent views of the models is important to improve the overall productivity. In this paper, we propose a collaborative model management framework for coordinating model change and automatic user-view update in a departmental computing environment. To do so, we describes changes in the model and their effects occurred in departmental model management environments and identifies the constructs and processes for maintaining the consistency between a shared model and its departmental user-views. Especially, in this framework, generic model concept was adopted for accommodating diverse mathematical models in a uniform way in a modelbase and object-oriented database management systems(ODBMS) for combining the model management constructs and automatic user-view update mechanisms in a single formalism. A prototype object-oriented modeling environment was developed using an ODBMS called ObjectStore and $C^{++}$ programming language on Windows NT.

• Journal of Institute of Control, Robotics and Systems
• /
• v.16 no.3
• /
• pp.269-276
• /
• 2010

In this paper, we present the detection method of moving objects based on two background models. These background models support to understand multi layered environment belonged in images taken by shaking camera and each model is MBM(Multiple Background Model) and TMBM (Temporal Median Background Model). Because two background models are Pixel-based model, it must have noise by camera movement. Therefore correlation coefficient calculates the similarity between consecutive images and measures camera motion vector which indicates camera movement. For the calculation of correlation coefficient, we choose the selected region and searching area in the current and previous image respectively then we have a displacement vector by the correlation process. Every selected region must have its own displacement vector therefore the global maximum of a histogram of displacement vectors is the camera motion vector between consecutive images. The MBM classifies the intensity distribution of each pixel continuously related by camera motion vector to the multi clusters. However, MBM has weak sensitivity for temporal intensity variation thus we use TMBM to support the weakness of system. In the video-based experiment, we verify the presented algorithm needs around 49(ms) to generate two background models and detect moving objects.

• Journal of Communications and Networks
• /
• v.12 no.5
• /
• pp.411-432
• /
• 2010

We consider three different secure broadcasting scenarios: i) Broadcast channels with common and confidential messages (BCC), ii) multi-receiver wiretap channels with public and confidential messages, and iii) compound wiretap channels. The BCC is a broadcast channel with two users, where in addition to the common message sent to both users, a private message, which needs to be kept hidden as much as possible from the other user, is sent to each user. In this model, each user treats the other user as an eavesdropper. The multi-receiver wiretap channel is a broadcast channel with two legitimate users and an external eavesdropper, where the transmitter sends a pair of public and confidential messages to each legitimate user. Although there is no secrecy concern about the public messages, the confidential messages need to be kept perfectly secret from the eavesdropper. The compound wiretap channel is a compound broadcast channel with a group of legitimate users and a group of eavesdroppers. In this model, the transmitter sends a common confidential message to the legitimate users, and this confidential message needs to be kept perfectly secret from all eavesdroppers. In this paper, we provide a survey of the existing information-theoretic results for these three forms of secure broadcasting problems, with a closer look at the Gaussian multiple-input multiple-output (MIMO) channel models. We also present the existing results for the more general discrete memoryless channel models, as they are often the first step in obtaining the capacity results for the corresponding Gaussian MIMO channel models.