• Journal of Institute of Control, Robotics and Systems
• /
• v.12 no.3
• /
• pp.201-207
• /
• 2006

A robust Kalman filtering method for uncertain stochastic systems is suggested by adopting a perturbation estimation process which is to reconstruct total uncertainty with respect to the nominal state transition equation. The predictor and corrector of discrete Kalman filter are reformulated with the perturbation estimator. Successively, the state and perturbation estimation error dynamics and the corresponding error covariance propagation equations are derived as well. Finally we have the recursive algorithm of Combined Kalman Filter-Perturbation Estimator (CKF). The proposed combined Kalman filter-perturbation estimator has the property of integrating innovations and the adaptation capability to system uncertainties. A numerical example is shown to demonstrate the effectiveness of the proposed scheme.

• Transactions of the Korean Society of Automotive Engineers
• /
• v.5 no.4
• /
• pp.100-109
• /
• 1997

Parameters in the internal combustion engines are variable depending on the operating points. Therefore, it is necessary to compensate for the uncertainties. Form this point of view, this paper gives a controller design method and a robust stability condition by LMI approach for engine torque control which satisfies the gives H$\infty$ control performance in the presence of physical parameter perturbations. To the end, the robustness of the system in the presence of perturbation is guaranteed in the all engine operating regions. Its effectiveness is demonstrated by simulation.

• Journal of Institute of Control, Robotics and Systems
• /
• v.17 no.9
• /
• pp.875-881
• /
• 2011

This paper describes a new method of estimating the gradient of a function with perturbation and correlation. We impose a known periodic perturbation to the input variable and observe the output of the function in order to obtain much richer and more reliable information. By taking the correlation between the input perturbation and the resultant function outputs, we can determine the gradient of the function. The computation of the correlation does not require derivatives; therefore the gradient can be estimated reliably. Robust estimation of the gradient using perturbation/correlation, which is very effective when an analytical solution is not available, is described. To verify the effectiveness of perturbation/correlation based estimation, the results of gradient estimation are compared with the analytical solutions of an example function. The effects of amplitude of the perturbation and number of samplings in a period are investigated. A minimization of a function with the gradient estimation method is performed.

• Proceedings of the KIEE Conference
• /
• 2006.04a
• /
• pp.219-221
• /
• 2006

The recently developed autopilot controller can make the input-output (I/O) dynamic characteristics of the nonlinear missile dynamics linear and independent of flight conditions. However, significant fin actuator dynamics can degenerate the I/O dynamic performance of the overall system. In this paper, we propose a singular perturbation-like method to compensate the effect of significant fin actuator dynamics in nonlinear missile control. The proposed compensation method does not require the time derivatives of fin angles but can maintain the linear I/O dynamic characteristics provided by the recently developed autopilot controller.

• Journal of Advanced Marine Engineering and Technology
• /
• v.17 no.5
• /
• pp.63-70
• /
• 1993

In 1980's to 1990's the marine propulsion diesel engines have been developed into lower speed and longer stroke for the energy saving(small S.F.O.C.). As these new trends the convetional mechnical-hydraulic governors were not adapted to the new requirements and the digital governors have been adopted in the marine use. The digital governors usually use the control algorithms such as the PID control, optimal control, adaptive control and etc. While the engine has delay time and parameter variations these control algorithms have difficulty in considering the stability and the robustness for the model uncertainty. In this study, the $H_{\infty}$ controller design method are applied to the speed control of the low speed marine diesel engine. By comparison the $H_{\infty}$ control results with the PID control results, the validity of the $H_{\infty}$ controller under the delay time and parameter variations is confirmed.

• The Transactions of the Korean Institute of Electrical Engineers
• /
• v.32 no.2
• /
• pp.43-49
• /
• 1983

In this study, model simplification problem using singular perturbation technique is considered. The correctness and errors of simplified model which is obtained by the use of this technique, depends upon the order and the time scaling factor of the simplified model But, unfortunately, there is no explicit criteria for selections of these parameters. In this paper, error equations are derived and expanded by using the useful properties of $L_2$-norm. Then, new criteria for selecting the order of the simplified model and time scaling factor with respect to error bound are suggested. Since these criteria, newly proposed in this study, have strong concern about error bound, it can be used to choose the minimum order of the simplified model and time scaling factor with respect to given error bound. Conversely, if the order of the simplified model and time scaling factor are given, the error induced by the simplification can also be computed easily.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2024.05a
• /
• pp.885-887
• /
• 2024

딥러닝 모델이 다양한 분야에 적용되면서, 딥러닝 모델에 대한 보안이 큰 이슈가 되고 있다. 특히, 입력 데이터에 섭동(perturbation)을 추가하여 모델의 정상적인 추론을 방해하는 적대적 공격(Adversarial Attack)에 대한 연구가 활발하게 진행되고 있다. 본 논문에서는 객체 탐지 모델의 NMS(Non-Maximum Suppression) 알고리즘에 대한 적대적 공격 기법 중 하나인 Phantom Sponges 공격을 수행하여 적대적 예제(Adversarial Example)를 생성하고, 원본 이미지와의 유사성을 측정하여 분석하고자 한다.

• Journal of the Institute of Electronics Engineers of Korea TC
• /
• v.40 no.6
• /
• pp.253-259
• /
• 2003

In this paper, the scattered field from a Pierson-Moskowitz sea surface assumed as the PEC by the Finite-Difference Time-Domain(FDTD) method was computed. A one-dimensional surface used to analysis scattering was generated by using the Pierson-Moskowitz model. Back scattering coefficients are calculated with different values of the wind speed(U) which determine configuration of the Pierson-Moskowitz sea surface. The number of surface realization for the computed field, the point number, and the width of surface realization are set to be 50, 8192, and 128k, respectively. In order to verify the computed values these results are compared with those of small perturbation methods, which show good agreement between them.

• Journal of the Korean Society for Aeronautical & Space Sciences
• /
• v.32 no.5
• /
• pp.58-64
• /
• 2004

A new method to obtain approximate solutions by placing the only poles of the slow subsystem for the two-time-scale aircraft dynamic systems. The two kinds of approximate solutions are obtained by a matrix block diagonalization. One is called the uncorrected solution, and the other is called the corrected solution. The former has an error of $O({\varepsilon})$, and the latter has an error of $O({\varepsilon}^2)$. Of course, both solutions are robust enough even though they are reduced solutions. The excellence of the proposed method is illustrated by an numerical example of an aircraft longitudinal dynamics.

• Journal of the Korean Society of Propulsion Engineers
• /
• v.24 no.4
• /
• pp.25-32
• /
• 2020

A numerical study on combustion instabilities in a model combustor was conducted with various momentum flux ratios. Five ratios are calculated based on an actual operating condition of rocket engine. As momentum flux ratio increases, the spreading angle on the injector outlet decreases. And, as increase of axial momentum flux, pressure fluctuation decreases inside the combustor. By using dynamic mode decomposition method, the acoustic modes inside the combustor are identified. Combustion stabilities are analyzed by comparing the damping coefficient of the 2^nd longitudinal mode.