• Journal of Advanced Navigation Technology
• /
• v.11 no.4
• /
• pp.371-377
• /
• 2007

Ionospheric time delay is one of the main error source for single-frequency DGPS applications, including time transfer and Wide Area Differential GPS (WADGPS). Grid-based algorithm was already developed for WADGPS but that algorithm is not applicable to geomagnetic storm condition in accuracy and management. In geomagnetic storm condition, the spatial distribution of vertical ionospheric delay is noisy and therefore the accuracy of modeling become low in grid-based algorithm. For better accuracy, function based algorithm can be used but the continuity of correction message is not guranteed. In this paper, we propose the ionospheric model using wavelet based algorithm. This algorithm shows better accuracy with the same number of correction message than the existing spherical harmonics algorithm and guarantees the continuity of correction messages when the number of message is expanded for geomagnetic storm condition.

• Smart Structures and Systems
• /
• v.22 no.1
• /
• pp.41-55
• /
• 2018

Damping ratio and frequency have influence on dynamic serviceability or instability such as vortex-induced vibration and displacement amplification due to earthquake and critical flutter velocity, and it is thus important to make determination of damping ratio and frequency accurate. As bridges are getting longer, small scale model test considering similitude law must be conducted to evaluate damping ratio and frequency. Analysis conditions modified by similitude law are applied to experimental test considering different scale ratios. Generally, Nyquist frequency condition based on natural frequency modified by similitude law has been used to determine sampling rate for different scale ratios, and total time length has been determined by users arbitrarily or by considering similitude law with respect to time for different scale ratios. However, Nyquist frequency condition is not suitable for multimode system with noisy signals. In addition, there is no specified criteria for determination of total time length. Those analysis conditions severely affect accuracy of damping ratio. The focus of this study is made on the determination of minimum analysis conditions for different scale ratios. Influence of signal to noise ratio is studied according to the level of noise level. Free initial value problem is proposed to resolve the condition that is difficult to know original initial value for free vibration. Ambient and free vibration tests were used to analyze the dynamic properties of a system using data collected from tests with a two degree-of-freedom section model and performed on full bridge 3D models of cable stayed bridges. The free decay is estimated with the stochastic subspace identification method that uses displacement data to measure damping ratios under noisy conditions, and the iterative least squares method that adopts low pass filtering and fourth order central differencing. Reasonable results were yielded in numerical and experimental tests.

• The Journal of the Acoustical Society of Korea
• /
• v.28 no.6
• /
• pp.566-571
• /
• 2009

This paper proposes noise-robust fast speaker adaptation method based on the eigenvoice framework in various noisy environments. The proposed method is focused on de-noising and environment clustering. Since the de-noised adaptation DB still has residual noise in itself, environment clustering divides the noisy adaptation data into similar environments by a clustering method using the cepstral mean of non-speech segments as a feature vector. Then each adaptation data in the same cluster is used to build an environment-clustered speaker adapted (SA) model. After selecting multiple environmentally clustered SA models which are similar to test environment, the speaker adaptation based on an appropriate linear combination of clustered SA models is conducted. According to our experiments, we observe that the proposed method provides error rate reduction of $40{\sim}59%$ over baseline with speaker independent model.

• Geophysics and Geophysical Exploration
• /
• v.10 no.1
• /
• pp.77-88
• /
• 2007

Following a successful bathymetric mapping demonstration in a previous study, the potential of airborne EM for seafloor characterisation has been investigated. The sediment thickness inferred from 1D inversion of helicopter-borne time-domain electromagnetic (TEM) data has been compared with estimates based on marine seismic studies. Generally, the two estimates of sediment thickness, and hence depth to resistive bedrock, were in reasonable agreement when the seawater was ${\sim}20\;m$ deep and the sediment was less than ${\sim}40\;m$ thick. Inversion of noisy synthetic data showed that recovered models closely resemble the true models, even when the starting model is dissimilar to the true model, in keeping with the uniqueness theorem for EM soundings. The standard deviations associated with shallow seawater depths inferred from noisy synthetic data are about ${\pm}5\;%$ of depth, comparable with the errors of approximately ${\pm}1\;m$ arising during inversion of real data. The corresponding uncertainty in depth-to-bedrock estimates, based on synthetic data inversion, is of order of ${\pm}10\;%$. The mean inverted depths of both seawater and sediment inferred from noisy synthetic data are accurate to ${\sim}1\;m$, illustrating the improvement in accuracy resulting from stacking. It is concluded that a carefully calibrated airborne TEM system has potential for surveying sediment thickness and bedrock topography, and for characterising seafloor resistivity in shallow coastal waters.

• The Journal of the Acoustical Society of Korea
• /
• v.40 no.3
• /
• pp.234-240
• /
• 2021

In this paper, the mask-based speech enhancement is improved for effective speech recognition in noise environments. In the mask-based speech enhancement, enhanced spectrum is obtained by multiplying the noisy speech spectrum by the mask. The VoiceFilter (VF) model is used as the mask estimation, and the Spectrogram Inpainting (SI) technique is used to remove residual noise of enhanced spectrum. In this paper, we propose a combined loss to further improve speech enhancement. In order to effectively remove the residual noise in the speech, the positive part of the Triplet loss is used with the component loss. For the experiment TIMIT database is re-constructed using NOISEX92 noise and background music samples with various Signal to Noise Ratio (SNR) conditions. Source to Distortion Ratio (SDR), Perceptual Evaluation of Speech Quality (PESQ), and Short-Time Objective Intelligibility (STOI) are used as the metrics of performance evaluation. When the VF was trained with the mean squared error and the SI model was trained with the combined loss, SDR, PESQ, and STOI were improved by 0.5, 0.06, and 0.002 respectively compared to the system trained only with the mean squared error.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.4
• /
• pp.387-393
• /
• 2002

The enhancement technique of noise signal using mixture HFM (Midden Filter Model) are proposed. Given the parameters of the clean signal and noise, noisy signal is modeled by a linear state-space model with Markov switching parameters. Estimation of state vector is required for estimating original signal. The estimation procedure is based on mixture interacting multiple model (MIMM) and the estimator of speech is given by the weighted sum of parallel Kalman filters operating interactively. Simulation results showed that the proposed method offers performance gains relative to the previous results with slightly increased complexity.

• Proceedings of the KIEE Conference
• /
• 2000.07d
• /
• pp.2299-2301
• /
• 2000

In this paper, the design of PID controller using Neural networks for the control of non-linear system is presented. First, non-linear system is identified using BPN(Backpropagation Network) algorithm. This identified model is connected to the PID controller and the parameters of PID controller are updated to the direction of reducing the difference between the identified model output and model reference output in arbitrary input signal. Therefore, identified model output tracks the model reference output in an acceptable error range and the parameters of controller are updated adaptively. The output of the system has a good performance in case of both noisy and noiseless model reference and we can control the system stable in off-line when the dynamics of the system is changed.

• Korean Journal of Remote Sensing
• /
• v.37 no.4
• /
• pp.719-731
• /
• 2021

This study proposes a two-stage hybrid classification model for crop classification using multi-temporal remote sensing images; the model combines feature embedding by using an autoencoder (AE) with a convolutional neural network (CNN) classifier to fully utilize features including informative temporal and spatial signatures. Long short-term memory (LSTM)-based AE (LAE) is fine-tuned using class label information to extract latent features that contain less noise and useful temporal signatures. The CNN classifier is then applied to effectively account for the spatial characteristics of the extracted latent features. A crop classification experiment with multi-temporal unmanned aerial vehicle images is conducted to illustrate the potential application of the proposed hybrid model. The classification performance of the proposed model is compared with various combinations of conventional deep learning models (CNN, LSTM, and convolutional LSTM) and different inputs (original multi-temporal images and features from stacked AE). From the crop classification experiment, the best classification accuracy was achieved by the proposed model that utilized the latent features by fine-tuned LAE as input for the CNN classifier. The latent features that contain useful temporal signatures and are less noisy could increase the class separability between crops with similar spectral signatures, thereby leading to superior classification accuracy. The experimental results demonstrate the importance of effective feature extraction and the potential of the proposed classification model for crop classification using multi-temporal remote sensing images.

• Structural Monitoring and Maintenance
• /
• v.7 no.3
• /
• pp.261-281
• /
• 2020

Civil structures may experience progressive deterioration and damage under environmental and operational conditions over their service life. Finite element (FE) model updating method is one of the most important approaches for damage identification in structures due to its capabilities in structural health monitoring. Although various damage detection approaches have been investigated on structures, there are limited studies on large-sized space structures. Thus, this paper aims to investigate the applicability and efficiency of sensitivity-based FE model updating framework for damage identification in large space structures from a distinct point of view. This framework facilitates modeling and model updating in large and geometric complicated space structures. Considering sensitivity-based FE model updating and vibration measurements, the discrepancy between acceleration response data in real damaged structure and hypothetical damaged structure have been minimized through adjusting the updating parameters. The feasibility and efficiency of the above-mentioned approach for damage identification has finally been demonstrated with two numerical examples: a flat double layer grid and a double layer diamatic dome. According to the results, this method can detect, localize, and quantify damages in large-scaled space structures very accurately which is robust to noisy data. Also, requiring a remarkably small number of iterations to converge, typically less than four, demonstrates the computational efficiency of this method.

• Journal of Korea Multimedia Society
• /
• v.24 no.2
• /
• pp.274-284
• /
• 2021

Customized services to a sleep induction for better sleepcare are more effective because of different satisfaction levels to users. The EOG data measured at the frontal lobe when a person blinks his eyes can be used as biometric data because it has different values for each person. The accuracy of measurement is degraded by a noise source, such as toss and turn. Therefore, it is necessary to analyze the noisy data and remove them from normal EOG by filtering. There are low-pass filtering and high-pass filtering as filtering using a frequency band. However, since filtering within a frequency band range is also required for more effective performance, we propose a machine learning model for the filtering of EOG data in this paper as the second filtering method. In addition, optimal values of parameters such as the depth of the hidden layer, the number of nodes of the hidden layer, the activation function, and the dropout were found through experiments, to improve the performance of the machine learning filtering model, and the filtering performance of 95.7% was obtained. Eventually, it is expected that it can be used for effective user identification services by using filtering model for EOG data.