• Proceedings of the KSR Conference
• /
• 2008.11b
• /
• pp.372-379
• /
• 2008

In the case of pantograph, three models of HSR350X(G7), KTX-I and KTX-II have been already introduced into the field of domestic high speed train. This thesis intends to explain performance test result of the conditionally purchasing pantograph that is progressing up to now. The pantograph is being developed to localize pantograph that was applied to KTX-I. Also, it consider criteria that applied for verification of design contents and method of dynamic test that verify pantograph's current collecting performance.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.17 no.4
• /
• pp.209-215
• /
• 2022

This paper proposes a novel approach for training an unsupervised depth estimation algorithm. The objective of unsupervised depth estimation is to estimate pixel-wise distances from camera without external supervision. While most previous works focus on model architectures, loss functions, and masking methods for considering dynamic objects, this paper focuses on the training framework to effectively use depth cue. The main loss function of unsupervised depth estimation algorithms is known as the photometric error. In this paper, we claim that direct depth cue is more effective than the photometric error. To obtain the direct depth cue, we adopt the technique of knowledge distillation which is a teacher-student learning framework. We train a teacher network based on a previous unsupervised method, and its depth predictions are utilized as pseudo labels. The pseudo labels are employed to train a student network. In experiments, our proposed algorithm shows a comparable performance with the state-of-the-art algorithm, and we demonstrate that our teacher-student framework is effective in the problem of unsupervised depth estimation.

• Journal of Information Processing Systems
• /
• v.20 no.2
• /
• pp.263-272
• /
• 2024

A model based on genetic algorithm optimization, GA-SVM, is proposed to warn university students of their status. This model improves the predictive effect of support vector machines. The genetic optimization algorithm is used to train the hyperparameters and adjust the kernel parameters, kernel penalty factor C, and gamma to optimize the support vector machine model, which can rapidly achieve convergence to obtain the optimal solution. The experimental model was trained on open-source datasets and validated through comparisons with random forest, backpropagation neural network, and GA-SVM models. The test results show that the genetic algorithm-optimized radial basis kernel support vector machine model GA-SVM can obtain higher accuracy rates when used for early warning in university learning.

• Journal of Korean Society of Transportation
• /
• v.25 no.2 s.95
• /
• pp.17-26
• /
• 2007

Because the prime objective of the current preemption methods at signalized intersections near highway-railroad grade crossings(IHRGCs) is to clear the crossing, secondary objectives such as safe pedestrian crossing time and minimized delay often are given less consideration or are ignored completely during the preemption. Under certain circumstances state-of-the-practice traffic signal preemption strategies may cause serious pedestrian safety and efficiency problems at IHRGCs. An improved transition preemption strategy(ITPS) that is specifically designed to improve intersection performance while maintaining or improving the current level of safety was developed by Cho and Rilett. Even if the new transition preemption strategy improved both the safety and efficiency of IHRGCs, the performance of the strategy is affected by train speed. Understanding the impact of this factor is essential in order to implement ITPS. In this paper, the effects of train speed were analyzed using a VISSIM simulation model which was calibrated to field conditions. It was concluded that the delay is affected more by train speed than the transitional preemption strategy and the safety of the intersection is not affected by train speed once an advanced preemption warning time(APWT) is equal to or greater than 90 seconds.

• Proceedings of the KSR Conference
• /
• 2011.10a
• /
• pp.2666-2671
• /
• 2011

In this paper, a theoretical approach was studied to make a baseline box type model satisfying the stiffness condition of a cut-out model. First, we compared the sum of the sectional theoretical deflections and the FEM result of the cut-out model under the static load test conditions, and we obtained good correlations from both the results. Second, To obtain the thickness of the baseline model, we used the mean value of geometric moment of intertia of the side wall and roof structure. Also, we compared the theoretical results and the FEM result of a baseline model, and we obtained good correlations. It is considered that the developed theoretical approach can be used for the weight reduction of train carbodies.

• Journal of the Korean Society of Safety
• /
• v.32 no.1
• /
• pp.115-120
• /
• 2017

The aim of this study is to develop construction safety and health management cost prediction model using support vector machine (SVM). To this end, theoretical concept of SVM is investigated to formulate the cost prediction model. Input and output variables have been selected by analyzing the balancing accounts for the completed construction project. In order to train and validate the proposed prediction model, 150 data sets have been gathered from field. Effects of SVM parameters on prediction accuracy are analyzed and from which the optimal parameter values have been determined. The prediction performance tests are conducted to confirm the applicability of the proposed model. Based on the results, it is concluded that the proposed SVM model can effectively be used to predict the construction safety and health management cost.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.24 no.2
• /
• pp.1-15
• /
• 1999

This research involves the development and evaluation of a work flow control model for a type of flexible manufacturing system(FMS) called a flexible flow line(FFL). The control model can be considered as a kind of hybrid intelligent model in that it utilizes both computer simulation and neural network technique. Training data sets were obtained using computer simulation of typical FFL states. And these data sets were used to train the neural network model. The model can easily incorporate particular aspects of a specific FFL such as limited buffer capacity and dispatching rules used. It also dynamically adapts to system uncertainty caused by such factors as machine breakdowns. Performance of the control model is shown to be superior to the random releasing method and the Minimal Part Set(MPS) heuristic in terms of machine utilization and work-in-process inventory level.

• Journal of the Korean Society for Precision Engineering
• /
• v.15 no.4
• /
• pp.15-26
• /
• 1998

Full car model is needed for investigating as a entire dynamics of vehicle. In this study, 7DOF of full car model's dynamics is selected. This paper proposes the output feedback controller based on optimal control theory. Input data and output data from the optimal controller are used for neural network system identification of the suspension system. To do system identification, neural network which has robustness against nonlinearities and disturbances is adapted. This study uses back-propagation algorithm to train a multil-layer neural network. After obtaining a neural network model of a suspension system, a neuro-controller is designed. Neuro-controller controls suspension system with off-line learning method and multistep ahead prediction model based on the neural network model and a neuro-controller. The optimal controller and the neuro-controller are designed and then, both performances are compared through. For simulation, sinusoidal and rectangular virtual bumps are selected.

• Journal of the Korean Society for Railway
• /
• v.10 no.4
• /
• pp.409-413
• /
• 2007

A dynamically scaled model of the catenary with a nominal scaling factor of 18.5:1 is designed and constructed. The motivation for developing such a model is the great difficulty of making accurate measurements on the full-scale catenary and the difficulty of making experimental modifications to it. The scaled model is designed to be dynamically equivalent to the full scale catenary with respect to the mass and elastic strength. The scaled model is partially verified by comparing linear vibration and wave characteristics with those predicted by the simulation study.

• International Journal of Internet, Broadcasting and Communication
• /
• v.15 no.2
• /
• pp.133-143
• /
• 2023

In this study, we propose a big data set of strawberry pests collected directly for diagnosis model learning and an automatic pest diagnosis model architecture based on deep learning. First, a big data set related to strawberry pests, which did not exist anywhere before, was directly collected from the web. A total of more than 12,000 image data was directly collected and classified, and this data was used to train a deep learning model. Second, the deep-learning-based automatic pest diagnosis module is a module that classifies what kind of pest or disease corresponds to when a user inputs a desired picture. In particular, we propose a model architecture that can optimally classify pests based on a convolutional neural network among deep learning models. Through this, farmers can easily identify diseases and pests without professional knowledge, and can respond quickly accordingly.