• Journal of Sensor Science and Technology
• /
• v.11 no.1
• /
• pp.54-59
• /
• 2002

This paper describes a new process technique for batch process of SOI(Si-on-Insulator) structures with buried cavities for MEMS(Micro Electro Mechanical System) applications by SDB(Si-wafer Direct Bonding) technology and electrochemical etch-stop. A low-cost electrochemical etch-stop method is used to control accurately the thickness of SOI. The cavities were made on the upper handling wafer by Si anisotropic etching. Two wafers are bonded with an intermediate insulating oxide layer. After high-temperature annealing($1000^{\circ}C$, 60 min), the SDB SOI structure with buried cavities was thinned by electrochemical etch-stop. The surface of the fabricated SDB SOI structure have more roughness that of lapping and polishing by mechanical method. This SDB SOI structure with buried cavities will provide a powerful and versatile substrate for novel microsensors arid microactuators.

• Proceedings of the Korean Institute of Navigation and Port Research Conference
• /
• v.1
• /
• pp.141-144
• /
• 2006

The deviations in the injection orbital parameters, resulting from launcher dispersions, need to be estimated and used for autonomous satellite operations. For the proposed small satellite mission of the university there will be two GPS receivers onboard the satellite to provide the instantaneous orbital state to the onboard data handling system. In order to meet the power requirements, the satellite will be sun-tracking whenever there is no imaging operation. For imaging activities, the satellite will be maneuvered to nadir-pointing mode. Due to such different modes of orientation the geometry for the GPS receivers will not be favorable at all times and there will be instances of poor geometry resulting in no output from the GPS receivers. Onboard the satellite, the orbital information should be continuously available for autonomous switching on/off of various subsystems. The paper presents the strategies to make use of small arcs of data from GPS receivers to compute the mean orbital parameters and use the updated orbital parameters to calculate the position and velocity whenever the same is not available from GPS receiver. Thus the navigation message from the GPS receiver, namely the position vector in Earth-Centered-Earth-Fixed (ECEF) frame, is used as measurements. As for estimation, two techniques - (1) batch least squares method, and (2) Kalman Filter method are used for orbit estimation (in real time). The performance of the onboard orbit estimation has been assessed based on hardware based multi-channel GPS Signal simulator. The results indicate good converge even with short arcs of data as the GPS navigation data are generally very accurate and the data rate is also fast (typically 1Hz).

• Journal of KIISE:Software and Applications
• /
• v.31 no.7
• /
• pp.949-958
• /
• 2004

race detection using a boosting based algorithm requires a very large size of face and nonface data. In addition, the fact that there always occurs a need for adding additional training data for better detection rates demands an efficient incremental teaming algorithm. In the design of incremental teaming based classifiers, the final classifier should represent the characteristics of the entire training dataset. Conventional methods have a critical problem in combining intermediate classifiers that weight updates depend solely on the performance of individual dataset. In this paper, for the purpose of application to face detection, we present a new method to combine an intermediate classifier with previously acquired ones in an optimal manner. Our algorithm creates a validation set by incrementally adding sampled instances from each dataset to represent the entire training data. The weight of each classifier is determined based on its performance on the validation set. This approach guarantees that the resulting final classifier is teamed by the entire training dataset. Experimental results show that the classifier trained by the proposed algorithm performs better than by AdaBoost which operates in batch mode, as well as by ${Learn}^{++}$.

• Journal of the Institute of Convergence Signal Processing
• /
• v.11 no.4
• /
• pp.320-325
• /
• 2010

This paper is about the method and procedure to register the candidate sequentially in the face recognition system using the PCA(Principal Components Analysis). We use the method to update the principal components sequentially with the SKL algorithm which is improved R-SVD algorithm. This algorithm enable us to solve the re-training problem of the increase the candidates number sequentially in the face recognition using the PCA. Also this algorithm can use in robust tracking system with the bright change based to the principal components. This paper proposes the procedure in the face recognition system which sequentially updates the principal components using the SKL algorithm. Then we compared the face recognition performance with the batch procedure for calculating the principal components using the standard KL algorithm and confirms the effects of the forgetting factor in the SKL algorithm experimentally.

• Journal of Soil and Groundwater Environment
• /
• v.19 no.1
• /
• pp.46-53
• /
• 2014

The purpose of this study is to promote utilization of paper mill sludge as an adsorbent for stabilizing heavy metals in contaminated water by measuring the adsorption capacity of paper mill sludge for cadmium and arsenic. To measure adsorption capacity of paper mill sludge, sorption isotherm experiments were analyzed by Langmuir and Freundlich isotherm models. Also, two methods of chemical modifications were applied to improve the adsorption capacities of paper-mill-sludge: the first method used sodium hydroxide (NaOH), called PMS-1, and the second method used the NaOH and tartaric acid ($C_4H_6O_6$) together, called PMS-2. For Cd adsorption, PMS-1 presented the increase of reactivity while PMS-2 presented the decline of reactivity compared to that of untreated paper-mill-sludge. In case of As adsorption, both PMS-1 and PMS-2 showed the decrease of adsorption capacities. This is because zeta-potential of paper mill sludge was changed to more negative values during chemical modification process due to the hydroxyl group in NaOH and the carboxyl group in $C_4H_6O_6$, respectively. Therefore, we may conclude that the chemical treatment process increases adsorption capacity of paper mill sludge for cation heavy metals such as Cd but not for As.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.37 no.1
• /
• pp.31-37
• /
• 2013

This study aims to find the optimal skirt dimensions for a composite pressure vessel with a separated dome part. The size optimization for the skirt structure of the composite pressure vessel was conducted using a sub-problem approximation method and batch processing codes programmed using ANSYS Parametric Design Language (APDL). The thickness and length of the skirt part were selected as design variables for the optimum analysis. The objective function and constraints were chosen as the weight and the displacement of the skirt part, respectively. The numerical results showed that the weight of the skirt of a composite pressure vessel with a separated dome part could be reduced by a maximum of 4.38% through size optimization analysis of the skirt structure.

• Journal of Soil and Groundwater Environment
• /
• v.18 no.5
• /
• pp.1-6
• /
• 2013

This study was performed to develop a method for quantitative analysis obtaining the amount of calcium carbonate minerals formed when Ca salts biomimetically reacted with carbon dioxide. There were two methods compared; 1) volumetric calcimeter method that determining the amount of released carbon dioxide after calcium carbonate minerals were acidified by 4N HCl and 2) Thermogravimetry-Differential Thermal Analysis (TG-DTA) adopting differential decomposition temperature breaking-up the structural link within calcium carbonate minerals. The comparisons were made by batch experiment (i.e., biocalcification process) along with control (i.e., nominal concentration of $CaCO_3$ prepared). For the control, TG-DTA took a minor root mean square deviation (RMSD) of 1.1~5.9 mg, whereas volumetric calcimeter exposed a greater RMSD of 28.3 mg. For the biocalcification, the amount of $CaCO_3$ was more precisely obtained for TG-DTA rather than that of volumetric calcimeter. It was decided that TG-DTA was more successfully used for quantitative analysis to observe the amount of calcium carbonate minerals derived from biocalcification.

• Korean Chemical Engineering Research
• /
• v.59 no.4
• /
• pp.532-541
• /
• 2021

Recently, research on the application of artificial intelligence in the chemical process has been increasing rapidly. However, overfitting is a significant problem that prevents the model from being generalized well to predict unseen data on test data, as well as observed training data. Cross validation is one of the ways to solve the overfitting problem. In this study, the time-series cross validation method was applied to optimize the number of batch and epoch in the hyperparameters of the prediction model for the 2,3-BDO distillation process, and it compared with K-fold cross validation generally used. As a result, the RMSE of the model with time-series cross validation was lower by 9.06%, and the MAPE was higher by 0.61% than the model with K-fold cross validation. Also, the calculation time was 198.29 sec less than the K-fold cross validation method.

• Nuclear Engineering and Technology
• /
• v.54 no.4
• /
• pp.1206-1212
• /
• 2022

Cold-crucible induction melting (CCIM) technology has been intensively studied as an advanced vitrification process for the immobilization of highly radioactive waste. This technology uses high-frequency induction to melt a glass matrix and waste, while the outer surface of the crucible is water-cooled, resulting in the formation of a frozen glass layer (skull). In this study, for the fabrication of borosilicate glass waste form, CCIM operation test with 60 kg of glass per batch was conducted using surrogate wastes composed of Cs, Sr, and Nd as a representative of highly radioactive nuclides generated during spent nuclear fuel management. A 60 kg-scale glass waste form was successfully fabricated through melting and draining processes using a CCIM system, and its physicochemical properties were analyzed. In particular, to enhance the controllability and reliability of the draining process, an air-cooling drain control method that can control draining through air-cooling near drain holes was developed, and its validity for draining control was verified. The method can offer controllability on various draining processes, such as molten salt or molten metal draining processes, and can be applied to a process requiring high throughput draining.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.12
• /
• pp.3904-3922
• /
• 2022

As a research hotspot, pedestrian detection has a wide range of applications in the field of computer vision in recent years. However, current pedestrian detection methods have problems such as insufficient detection accuracy and large models that are not suitable for large-scale deployment. In view of these problems mentioned above, a lightweight pedestrian detection and early warning method using a new model called you only look once (Yolov5) is proposed in this paper, which utilizing advantages of Yolov5s model to achieve accurate and fast pedestrian recognition. In addition, this paper also optimizes the loss function of the batch normalization (BN) layer. After sparsification, pruning and fine-tuning, got a lot of optimization, the size of the model on the edge of the computing power is lower equipment can be deployed. Finally, from the experimental data presented in this paper, under the training of the road pedestrian dataset that we collected and processed independently, the Yolov5s model has certain advantages in terms of precision and other indicators compared with traditional single shot multiBox detector (SSD) model and fast region-convolutional neural network (Fast R-CNN) model. After pruning and lightweight, the size of training model is greatly reduced without a significant reduction in accuracy, and the final precision reaches 87%, while the model size is reduced to 7,723 KB.