• Journal of the Korea Society of Computer and Information
• /
• v.28 no.4
• /
• pp.1-11
• /
• 2023

In this paper, instead of using recurrent neural network, we compare a classification performance of time series imaging algorithms using convolution neural network. There are traditional algorithms that imaging time series data (e.g. GAF(Gramian Angular Field), MTF(Markov Transition Field), RP(Recurrence Plot)) in TSC(Time Series Classification) community. Furthermore, we compare STFT(Short Time Fourier Transform) algorithm that can acquire spectrogram that visualize feature of voice data. We experiment CNN's performance by adjusting hyper parameters of imaging algorithms. When evaluate with GunPoint dataset in UCR archive, STFT(Short-Time Fourier transform) has higher accuracy than other algorithms. GAF has 98~99% accuracy either, but there is a disadvantage that size of image is massive.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.57-58
• /
• 2022

Sleep apnea (SA) is a common chronic sleep disorder that disrupts breathing during sleep. Clinically, the standard for diagnosing SA involves nocturnal polysomnography (PSG). However, this requires expert human intervention and considerable time, which limits the availability of SA diagnoses in public health sectors. Therefore, ECG-based methods for SA detection have been proposed to automate the PSG procedure and reduce its discomfort. We propose a preprocessing method to convert the one-dimensional time series of ECG into two-dimensional images using the Gramian Angular Field (GAF) algorithm, extract temporal features, and use a two-dimensional convolutional neural network for classification. The results of this study demonstrated that the proposed method can perform SA detection with specificity, sensitivity, accuracy, and area under the curve (AUC) of 88.89%, 81.50%, 86.11%, and 0.85, respectively. Our experimental results show that SA is successfully classified by extracting preprocessing transforms with temporal features.

• Journal of IKEEE
• /
• v.28 no.2
• /
• pp.180-186
• /
• 2024

This paper proposes an efficient fault diagnosis method for ITSC(Inter-Turn Short Circuit) in three-phase induction motors using CNN. By utilizing only the D-axis component of the D-Q synchronous coordinate system, it compares SWM(Slide Window Method) and GAF(Gramian Angular Field) methods for image encoding. Results show GAF achieving ~74% accuracy, while SWM achieves ~65%, indicating GAF's superiority by 9%. Learning time (~14.74s) remains consistent, particularly with epochs ≤ 100, showcasing faster learning.

• 제어로봇시스템학회:학술대회논문집
• /
• 1997.10a
• /
• pp.1452-1455
• /
• 1997

This paper studies the problem of approximating commensurate input delay sustems by finite dimensional systems based on the Hankel singular values. I is shown that the Gankel singular values are solutions a trancendental equation and the Hankel singular vectors are obtained form the kernel of the matrix. The computaioin is carried out in state spae framework. Once singular values and vectors are calcualted, finite dimensional approximated systems are constructed using stadnard linear system computational tools. An example is included.

• 제어로봇시스템학회:학술대회논문집
• /
• 1991.10a
• /
• pp.1219-1219
• /
• 1991

This paper proposes the model order reduction with pass band error reduction in the frequency domain of discrete time linear systems. The algorithm is the new method of reduced order model which reduces passband error by changing controllability and observability gramian used with weighted functions. A numerical example shows that this algorithm has lower passband error than balanced w&l and weighted function characteristics in frequency domain.

• Proceedings of the Korean Institute of Communication Sciences Conference
• /
• 1991.10a
• /
• pp.215-219
• /
• 1991

This paper is equivalent model transform which reduces the restriction of digitalization in the discrete linear system. This algorithm is the method that weight is given to contribillity and obserbility gramian, the regular matrix T of coordinate transform is obtained and then the state space coefficents of weighted model can be obtained. This study shows the frequency reponse of low quantization error according to the order of weighting function. The result shows that frequency response of the proposed algorithm is better than that of the balanced realization in the system of smaller bit.

• Journal of Internet Computing and Services
• /
• v.23 no.4
• /
• pp.11-19
• /
• 2022

This paper is a study to evaluate the performance when classifying explosive components by capacity using a convolutional neural network (CNN). Among the existing explosive classification methods, the IMS steam detector method determines the presence or absence of an explosive only when the explosive concentration exceeds the threshold set by the user. The IMS steam detector has a problem of determining that even if an explosive exists, the explosive does not exist in an amount that does not exceed the threshold. Therefore, it is necessary to detect the explosive component even when the concentration of the explosive component does not exceed the threshold. Accordingly, in this paper, after imaging explosive time series data with the Gramian Angular Field (GAF) algorithm, it is possible to determine whether there are explosive components and the amount of explosive components even when the concentration of explosive components does not exceed a threshold.

• 제어로봇시스템학회:학술대회논문집
• /
• 1997.10a
• /
• pp.1651-1654
• /
• 1997

Restiction of vertical and lateral accelerations is one of the very important requierments which has to be satisfied on the practice of automatically controlled flights of the civil aviation passenger planes. This goal could be achived on the basis of the optimization procedure using specilly constructed quadratic performance index. In the report the application of this procedure to the parameteric optimization of the control laws with known structure for autopilot of midium-size aircraft in the level flight model is demonstrated. Performance index is calculted on the basis of the controllability grammian. Results of simulation of control processes in the lateral and longitudinal channels sre represented.

• Journal of Institute of Control, Robotics and Systems
• /
• v.8 no.5
• /
• pp.351-356
• /
• 2002

This papaer deals with a model reduction problem for linear systems with time varying parameters. For this problem, a controllability Grammian and an observability Grammian are introduced and computed by solving linear matrix inequalities. Using the controllability/observability Grammian, a balanced state space realization for linear parameter varying systems is obtained. From the balanced state space realization, a reduced model can be obtained by truncating not only states but also time varying parameters and an upper bound of the model reduction error is derived as well.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 1996.04a
• /
• pp.299-304
• /
• 1996

This paper studies the rational approximation of multiple input delay systems using the Hankel singular values and vectors, which are the soultion of a transcendental equation. Rational approximatants are obtained from output normal realizations which are constructed by the Hankel singular values and vectors. Consequently, it is shown that rational approximants by output normal realization preserve intrinsic properties of time delay systems than Pade approximants.