• Journal of Navigation and Port Research
• /
• v.46 no.4
• /
• pp.367-374
• /
• 2022

The purpose of this study was to propose a deep learning algorithm that applies to the fault diagnosis of fuel pumps and purifiers of autonomous ships. A deep learning algorithm reflecting the time dependence of the measured signal was configured, and the failure pattern was trained using the vibration signal, measured in the equipment's regular operation and failure state. Considering the sequential time-dependence of deterioration implied in the vibration signal, this study adopts Conv1D with sliding window computation for fault detection. The time dependence was also reflected, by transferring the measured signal from two-dimensional to three-dimensional. Additionally, the optimal values of the hyper-parameters of the Conv1D model were determined, using the grid search technique. Finally, the results show that the proposed data preprocessing method as well as the Conv1D model, can reflect the sequential dependency between the fault and its effect on the measured signal, and appropriately perform anomaly as well as failure detection, of the equipment chosen for application.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.50 no.4
• /
• pp.287-295
• /
• 2022

As low altitude long-range artillery threat has been strengthened, the development of anti-artillery interception system to protect assets against its attacks will be kicked off. We view the defense of long-range artillery attacks as a typical dynamic weapon target assignment (DWTA) problem. DWTA is a sequential decision process in which decision making under future uncertain attacks affects the subsequent decision processes and its results. These are typical characteristics of Markov decision process (MDP) model. We formulate the problem as a MDP model to examine the assignment policy for the defender. The proximity of the capital of South Korea to North Korea border limits the computation time for its solution to a few second. Within the allowed time interval, it is impossible to compute the exact optimal solution. We apply approximate dynamic programming (ADP) approach to check if ADP approach solve the MDP model within processing time limit. We employ Shoot-Shoot-Look policy as a baseline strategy and compare it with ADP approach for three scenarios. Simulation results show that ADP approach provide better solution than the baseline strategy.

• Smart Media Journal
• /
• v.12 no.1
• /
• pp.9-16
• /
• 2023

Recently, a convolutional neural network (CNN) based system is being developed to overcome the limitations of human resources in the apple quality classification of farmhouse. However, since convolutional neural networks receive only images of the same size, preprocessing such as sampling may be required, and in the case of oversampling, information loss of the original image such as image quality degradation and blurring occurs. In this paper, in order to minimize the above problem, to generate a image patch based graph of an original image and propose a random walk-based positional encoding method to apply the graph transformer model. The above method continuously learns the position embedding information of patches which don't have a positional information based on the random walk algorithm, and finds the optimal graph structure by aggregating useful node information through the self-attention technique of graph transformer model. Therefore, it is robust and shows good performance even in a new graph structure of random node order and an arbitrary graph structure according to the location of an object in an image. As a result, when experimented with 5 apple quality datasets, the learning accuracy was higher than other GNN models by a minimum of 1.3% to a maximum of 4.7%, and the number of parameters was 3.59M, which was about 15% less than the 23.52M of the ResNet18 model. Therefore, it shows fast reasoning speed according to the reduction of the amount of computation and proves the effect.

• Journal of Radiation Protection and Research
• /
• v.28 no.3
• /
• pp.199-206
• /
• 2003

Because the MIRD phantom, the representative mathematical phantom was developed for the calculation of internal radiation dose, and simulated by the simplified mathematical equations for rapid computation, the appropriateness of application to external dose calculation and the closeness to real human body should be justified. This study was intended to modify the MIRD phantom according to the comparison of the organ absorbed doses in the two phantoms exposed to monoenergetic broad parallel photon beams of the energy between 0.05 MeV and 10 MeV. The organ absorbed doses of the MIRD phantom and the Zubal yokel phantom were calculated for AP and PA geometries by MCNP4C, general-purpose Monte Carlo code. The MIRD phantom received higher doses than the Zubal phantom for both AP and PA geometries. Effective dose in PA geometry for 0.05 MeV photon beams showed the difference up to 50%. Anatomical axial views of the two phantoms revealed the thinner trunk thickness of the MIRD phantom than that of the Zubal phantom. To find out the optimal thickness of trunk, the difference of effective doses for 0.5 MeV photon beams for various trunk thickness of the MIRD phantom from 20 cm to 36 cm were compared. The optimal thunk thickness, 24 cm and 28 cm for AP and PA geometries, respectively, showed the minimum difference of effective doses between the two phantoms. The trunk model of the MIRD phantom was modified and the organ doses were recalculated using the modified MIRD phantom. The differences of effective dose for AP and PA geometries reduced to 7.3% and the overestimation of organ doses decreased, too. Because MIRD-type phantoms are easier to be adopted in Monte Carlo calculations and to standardize, the modifications of the MIRD phantom allow us to hold the advantage of MIRD-type phantoms over a voxel phantom and alleviate the anatomical difference and consequent disagreement in dose calculation.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.37 no.1
• /
• pp.67-76
• /
• 2024

In this paper, we present a DIP-MLS testing method that combines digital image processing with a rigid body-based MLS differencing approach to measure mechanical variables and analyze the impact of target location and image resolution. This method assesses the displacement of the target attached to the sample through digital image processing and allocates this displacement to the node displacement of the MLS differencing method, which solely employs nodes to calculate mechanical variables such as stress and strain of the studied object. We propose an effective method to measure the displacement of the target's center of gravity using digital image processing. The calculation of mechanical variables through the MLS differencing method, incorporating image-based target displacement, facilitates easy computation of mechanical variables at arbitrary positions without constraints from meshes or grids. This is achieved by acquiring the accurate displacement history of the test specimen and utilizing the displacement of tracking points with low rigidity. The developed testing method was validated by comparing the measurement results of the sensor with those of the DIP-MLS testing method in a three-point bending test of a rubber beam. Additionally, numerical analysis results simulated only by the MLS differencing method were compared, confirming that the developed method accurately reproduces the actual test and shows good agreement with numerical analysis results before significant deformation. Furthermore, we analyzed the effects of boundary points by applying 46 tracking points, including corner points, to the DIP-MLS testing method. This was compared with using only the internal points of the target, determining the optimal image resolution for this testing method. Through this, we demonstrated that the developed method efficiently addresses the limitations of direct experiments or existing mesh-based simulations. It also suggests that digitalization of the experimental-simulation process is achievable to a considerable extent.

• Journal of Intelligence and Information Systems
• /
• v.24 no.1
• /
• pp.205-225
• /
• 2018

Convolutional Neural Network (ConvNet) is one class of the powerful Deep Neural Network that can analyze and learn hierarchies of visual features. Originally, first neural network (Neocognitron) was introduced in the 80s. At that time, the neural network was not broadly used in both industry and academic field by cause of large-scale dataset shortage and low computational power. However, after a few decades later in 2012, Krizhevsky made a breakthrough on ILSVRC-12 visual recognition competition using Convolutional Neural Network. That breakthrough revived people interest in the neural network. The success of Convolutional Neural Network is achieved with two main factors. First of them is the emergence of advanced hardware (GPUs) for sufficient parallel computation. Second is the availability of large-scale datasets such as ImageNet (ILSVRC) dataset for training. Unfortunately, many new domains are bottlenecked by these factors. For most domains, it is difficult and requires lots of effort to gather large-scale dataset to train a ConvNet. Moreover, even if we have a large-scale dataset, training ConvNet from scratch is required expensive resource and time-consuming. These two obstacles can be solved by using transfer learning. Transfer learning is a method for transferring the knowledge from a source domain to new domain. There are two major Transfer learning cases. First one is ConvNet as fixed feature extractor, and the second one is Fine-tune the ConvNet on a new dataset. In the first case, using pre-trained ConvNet (such as on ImageNet) to compute feed-forward activations of the image into the ConvNet and extract activation features from specific layers. In the second case, replacing and retraining the ConvNet classifier on the new dataset, then fine-tune the weights of the pre-trained network with the backpropagation. In this paper, we focus on using multiple ConvNet layers as a fixed feature extractor only. However, applying features with high dimensional complexity that is directly extracted from multiple ConvNet layers is still a challenging problem. We observe that features extracted from multiple ConvNet layers address the different characteristics of the image which means better representation could be obtained by finding the optimal combination of multiple ConvNet layers. Based on that observation, we propose to employ multiple ConvNet layer representations for transfer learning instead of a single ConvNet layer representation. Overall, our primary pipeline has three steps. Firstly, images from target task are given as input to ConvNet, then that image will be feed-forwarded into pre-trained AlexNet, and the activation features from three fully connected convolutional layers are extracted. Secondly, activation features of three ConvNet layers are concatenated to obtain multiple ConvNet layers representation because it will gain more information about an image. When three fully connected layer features concatenated, the occurring image representation would have 9192 (4096+4096+1000) dimension features. However, features extracted from multiple ConvNet layers are redundant and noisy since they are extracted from the same ConvNet. Thus, a third step, we will use Principal Component Analysis (PCA) to select salient features before the training phase. When salient features are obtained, the classifier can classify image more accurately, and the performance of transfer learning can be improved. To evaluate proposed method, experiments are conducted in three standard datasets (Caltech-256, VOC07, and SUN397) to compare multiple ConvNet layer representations against single ConvNet layer representation by using PCA for feature selection and dimension reduction. Our experiments demonstrated the importance of feature selection for multiple ConvNet layer representation. Moreover, our proposed approach achieved 75.6% accuracy compared to 73.9% accuracy achieved by FC7 layer on the Caltech-256 dataset, 73.1% accuracy compared to 69.2% accuracy achieved by FC8 layer on the VOC07 dataset, 52.2% accuracy compared to 48.7% accuracy achieved by FC7 layer on the SUN397 dataset. We also showed that our proposed approach achieved superior performance, 2.8%, 2.1% and 3.1% accuracy improvement on Caltech-256, VOC07, and SUN397 dataset respectively compare to existing work.