• Korean Journal of Remote Sensing
• /
• v.38 no.5_1
• /
• pp.447-460
• /
• 2022

It is well known that satellite synthetic aperture radar interferometry (InSAR) has been widely used for the observation of surface displacement owing to earthquakes, volcanoes, and subsidence very precisely. In wetlands where vegetation exists on the surface of the water, it is possible to create a water level change map with high spatial resolution over a wide area using the InSAR technique. Currently, a number of imaging radar satellites are in operation, and most of them support a ScanSAR mode observation to gather information over a large area at once. The Cienaga Grande de Santa Marta (CGSM) wetland, located in northern Colombia, is a vast wetland developed along the Caribbean coast. The CGSM wetlands face serious environmental threats from human activities such as reclamation for agricultural uses and residential purposes as well as natural causes such as sea level rise owing to climate change. Various restoration and protection plans have been conducted to conserve these invaluable environments in recognition of the ecological importance of the CGSM wetlands. Monitoring of water level changes in wetland is very important resources to understand the hydrologic characteristics and the in-situ water level gauge stations are usually utilized to measure the water level. Although it can provide very good temporal resolution of water level information, it is limited to fully understand flow pattern owing to its very coarse spatial resolution. In this study, we evaluate the L-band ALOS-2 PALSAR-2 ScanSAR mode to observe the water level change over the wide wetland area using the radar interferometric technique. In order to assess the quality of the interferometric product in the aspect of spatial resolution and coherence, we also utilized ALOS-2 PALSAR-2 stripmap high-resolution mode observations.

• Journal of the Korean Wood Science and Technology
• /
• v.13 no.1
• /
• pp.19-62
• /
• 1985

This study was performed to investigate: (i) the bending processing properties of silk worm oak (Quercus acutissima Carr.) and Korean red pine (Pinus densiflora S. et Z.) by boiling and steaming treatments; (ii) the effects of interrelated factors - sapwood and heartwood, annual ring placement, softening temperature and time, moisture content. and wood defects on bending processing properties; (iii) the changing rates of bending radii after release from a tension strap, and (iv) the improving methods of bending process by treatment with chemicals. The size of specimens tested was $15{\times}15{\times}350mm$ for boiling and steaming treatments and $5{\times}10{\times}200mm$ for treatments with chemicals. The specimens were green for boiling treatments and dried to 15 percent for steaming treatments. The specimens for treatments with chemicals were soaked in saturated urea solution, 35 percent formaldehyde solution, 25 percent polyethylene glycol -400 solution, and 25 percent ammonium hydroxide solution for 5 days and immediately followed the bending process, respectively. The results obtained were as follows: 1. The internal temperature of silk worm oak and Korean red pine by boiling and steaming time was raised slowly to $30^{\circ}C$ but rapidly from $30^{\circ}C$ to $80-90^{\circ}C$ and then slowly from $80-90^{\circ}C$ to $100^{\circ}C$. 2. The softening time required to the final temperature was directly proportional to the thickness of specimen. The time required from $25^{\circ}C$ to $100^{\circ}C$ for 15mm-squared specimen was 9.6-11.2 minutes in silk worm oak and 7.6-8.1 minutes in Korean red pine. 3. The moisture content (M.C.) of specimen by steaming time was increased rapidly first 4 minutes in the both species, and moderately from 4 to 20 minutes and then slowly and constantly in silk worm oak, and moderately from 4 to 15 minutes and then slowly and constantly in Korean red pine. The M.C. of 15mm-squared specimen in 50 minutes of steaming was increased to 18.0 percent in the oak and 22.4 percent in the pine from the initial conditioned M.C. of 15 percent The rate of moisture adsorption measured was therefore faster in the pine than in the oak. 4. The mechanical properties of the both species were decreased significantly with the increase of boiling rime. The decrement by the boiling treatment for 60 minutes was measured to 36.6-45.0 percent in compressive strength, 12.5-17.5 percent in tensile strength, 31.6-40.9 percent in modulus of rupture, and 23.3-34.6 percent in modulus of elasticity. 5. The minimum bending radius (M.B.R.) of sapwood and heartwood was 60-80 mm and 90 mm in silk worm oak, and 260 - 300 mm and 280 - 300 mm in Korean red pine, respectively. Therefore, the both species showed better bending processing properties in sapwood than in heartwood. 6. The M.B.R. of edge-grained and flat-grained specimen in suk worm oak was 60-80 mm, but the M.B.R. in Korean red pine was 240-280 mm and 260-360 mm, respectively. Comparing the M.B.R. of edge-grained with flat-grained specimen, in the pine the edge-grained showed better bending processing property than the flat-grained. 7. The bending processing properties of the both species were improved by the rising of softening temperature from $40^{\circ}C$ to $100^{\circ}C$. The minimum softening temperature for bending was $90^{\circ}C$ in silk worm oak and $80^{\circ}C$ in Korean red pine, and the dependency of softening temperature for bending was therefore higher in the oak than in the pine. 8. The bending processing properties of the both species were improved by the increase of softening time as well as temperature, but even after the internal temperature of specimen reaching to the final temperature, somewhat prolonged softening was required to obtain the best plastic conditions. The minimum softening time for bending of 15 mm-squared silk worm oak and Korean red pine specimen was 15 and 10 minutes in the boiling treatment, and 30 and 20 minutes in the steaming treatment, respectively. 9. The optimum M.C. for bending of silk worm oak was 20 percent, and the M.C. above fiber saturation point rather degraded the bending processing property, whereas the optimum M.C. of Korean red pine needed to be above 30 percent. 10. The bending works in the optimum conditions obtained as seen in Table 24 showed that the M.B.R. of silk worm oak and Korean red pine was 80 mm and 240 mm in the boiling treatment, and 50 mm and 280 mm in the steaming treatment, respectively. Therefore, the bending processing property of the oak was better in the steaming than in the boiling treatment, but that of the pine better in the boiling than in the steaming treatment. 11. In the bending without a tension strap, the radio r/t of the minimum bending radius t to the thickness t of silk worm oak and Korean red pine specimen amounted to 16.0 and 21.3 in the boiling treatment, and 17.3 and 24.0 in the steaming treatment, respectively. But in the bending with a tension strap, the r/t of the oak and the pine specimen decreased to 5.3 and 16.0 in t he boiling treatment, and 3.3 and 18.7 in the steaming treatment, respectively. Therefore, the bending processing properties of the both species were significantly improved by the strap. 12. The effect of pin knot on the degradation of bending processing property was very severe in silk worm oak by side, e.g. 90 percent of the oak specimens with pin knot on the concave side were ruptured when bent to a 100 mm radius but only 10 percent of the other specimens with pin knot on the convex side were ruptured. 13. The changing rate in the bending radius of specimen bent to a 300 mm radius after 30 days of exposure to room temperature conditions was measured to 4.0-10.3 percent in the boiling treatment and 13,0-15.0 percent in the steaming treatment. Therefore, the degree of spring back after release was higher in the steaming than in the boiling treatment. And the changing rate of moisture-proofing treated specimen by expoxy resin coating was only -1.0.0 percent. 14. Formaldehyde, 35 percent solution, and 25 percent polyethylene glycol-400 solution found no effect on the plasticization of the both species, but saturated urea solution and 25 percent ammonium hydroxide solution found significant effect in comparison to non-treated specimen. But the effect of the treatment with chemicals alone was inferior to that of the steaming treatment, and the steaming treatment after the treatment with chemicals improved 10-24 percent over the bending processing property of steam-bent specimen. 15. Three plasticity coefficients - load-strain coefficient, strain coefficient, and energy coefficient - were evaluated to be appropriate for the index of bending processing property because the coefficients had highly significant correlation with the bending radius. The fitness of the coefficients as the index was good at load-strain coefficient, energy coefficient, and strain coefficient, in order.

• Journal of Intelligence and Information Systems
• /
• v.23 no.2
• /
• pp.71-88
• /
• 2017

Language models were originally developed for speech recognition and language processing. Using a set of example sentences, a language model predicts the next word or character based on sequential input data. N-gram models have been widely used but this model cannot model the correlation between the input units efficiently since it is a probabilistic model which are based on the frequency of each unit in the training set. Recently, as the deep learning algorithm has been developed, a recurrent neural network (RNN) model and a long short-term memory (LSTM) model have been widely used for the neural language model (Ahn, 2016; Kim et al., 2016; Lee et al., 2016). These models can reflect dependency between the objects that are entered sequentially into the model (Gers and Schmidhuber, 2001; Mikolov et al., 2010; Sundermeyer et al., 2012). In order to learning the neural language model, texts need to be decomposed into words or morphemes. Since, however, a training set of sentences includes a huge number of words or morphemes in general, the size of dictionary is very large and so it increases model complexity. In addition, word-level or morpheme-level models are able to generate vocabularies only which are contained in the training set. Furthermore, with highly morphological languages such as Turkish, Hungarian, Russian, Finnish or Korean, morpheme analyzers have more chance to cause errors in decomposition process (Lankinen et al., 2016). Therefore, this paper proposes a phoneme-level language model for Korean language based on LSTM models. A phoneme such as a vowel or a consonant is the smallest unit that comprises Korean texts. We construct the language model using three or four LSTM layers. Each model was trained using Stochastic Gradient Algorithm and more advanced optimization algorithms such as Adagrad, RMSprop, Adadelta, Adam, Adamax, and Nadam. Simulation study was done with Old Testament texts using a deep learning package Keras based the Theano. After pre-processing the texts, the dataset included 74 of unique characters including vowels, consonants, and punctuation marks. Then we constructed an input vector with 20 consecutive characters and an output with a following 21st character. Finally, total 1,023,411 sets of input-output vectors were included in the dataset and we divided them into training, validation, testsets with proportion 70:15:15. All the simulation were conducted on a system equipped with an Intel Xeon CPU (16 cores) and a NVIDIA GeForce GTX 1080 GPU. We compared the loss function evaluated for the validation set, the perplexity evaluated for the test set, and the time to be taken for training each model. As a result, all the optimization algorithms but the stochastic gradient algorithm showed similar validation loss and perplexity, which are clearly superior to those of the stochastic gradient algorithm. The stochastic gradient algorithm took the longest time to be trained for both 3- and 4-LSTM models. On average, the 4-LSTM layer model took 69% longer training time than the 3-LSTM layer model. However, the validation loss and perplexity were not improved significantly or became even worse for specific conditions. On the other hand, when comparing the automatically generated sentences, the 4-LSTM layer model tended to generate the sentences which are closer to the natural language than the 3-LSTM model. Although there were slight differences in the completeness of the generated sentences between the models, the sentence generation performance was quite satisfactory in any simulation conditions: they generated only legitimate Korean letters and the use of postposition and the conjugation of verbs were almost perfect in the sense of grammar. The results of this study are expected to be widely used for the processing of Korean language in the field of language processing and speech recognition, which are the basis of artificial intelligence systems.

• Journal of Intelligence and Information Systems
• /
• v.26 no.4
• /
• pp.127-148
• /
• 2020

The data center is a physical environment facility for accommodating computer systems and related components, and is an essential foundation technology for next-generation core industries such as big data, smart factories, wearables, and smart homes. In particular, with the growth of cloud computing, the proportional expansion of the data center infrastructure is inevitable. Monitoring the health of these data center facilities is a way to maintain and manage the system and prevent failure. If a failure occurs in some elements of the facility, it may affect not only the relevant equipment but also other connected equipment, and may cause enormous damage. In particular, IT facilities are irregular due to interdependence and it is difficult to know the cause. In the previous study predicting failure in data center, failure was predicted by looking at a single server as a single state without assuming that the devices were mixed. Therefore, in this study, data center failures were classified into failures occurring inside the server (Outage A) and failures occurring outside the server (Outage B), and focused on analyzing complex failures occurring within the server. Server external failures include power, cooling, user errors, etc. Since such failures can be prevented in the early stages of data center facility construction, various solutions are being developed. On the other hand, the cause of the failure occurring in the server is difficult to determine, and adequate prevention has not yet been achieved. In particular, this is the reason why server failures do not occur singularly, cause other server failures, or receive something that causes failures from other servers. In other words, while the existing studies assumed that it was a single server that did not affect the servers and analyzed the failure, in this study, the failure occurred on the assumption that it had an effect between servers. In order to define the complex failure situation in the data center, failure history data for each equipment existing in the data center was used. There are four major failures considered in this study: Network Node Down, Server Down, Windows Activation Services Down, and Database Management System Service Down. The failures that occur for each device are sorted in chronological order, and when a failure occurs in a specific equipment, if a failure occurs in a specific equipment within 5 minutes from the time of occurrence, it is defined that the failure occurs simultaneously. After configuring the sequence for the devices that have failed at the same time, 5 devices that frequently occur simultaneously within the configured sequence were selected, and the case where the selected devices failed at the same time was confirmed through visualization. Since the server resource information collected for failure analysis is in units of time series and has flow, we used Long Short-term Memory (LSTM), a deep learning algorithm that can predict the next state through the previous state. In addition, unlike a single server, the Hierarchical Attention Network deep learning model structure was used in consideration of the fact that the level of multiple failures for each server is different. This algorithm is a method of increasing the prediction accuracy by giving weight to the server as the impact on the failure increases. The study began with defining the type of failure and selecting the analysis target. In the first experiment, the same collected data was assumed as a single server state and a multiple server state, and compared and analyzed. The second experiment improved the prediction accuracy in the case of a complex server by optimizing each server threshold. In the first experiment, which assumed each of a single server and multiple servers, in the case of a single server, it was predicted that three of the five servers did not have a failure even though the actual failure occurred. However, assuming multiple servers, all five servers were predicted to have failed. As a result of the experiment, the hypothesis that there is an effect between servers is proven. As a result of this study, it was confirmed that the prediction performance was superior when the multiple servers were assumed than when the single server was assumed. In particular, applying the Hierarchical Attention Network algorithm, assuming that the effects of each server will be different, played a role in improving the analysis effect. In addition, by applying a different threshold for each server, the prediction accuracy could be improved. This study showed that failures that are difficult to determine the cause can be predicted through historical data, and a model that can predict failures occurring in servers in data centers is presented. It is expected that the occurrence of disability can be prevented in advance using the results of this study.