• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.15 no.9
• /
• pp.3243-3257
• /
• 2021

Deep neural networks provide excellent performance in pattern recognition, audio classification, and image recognition. It is important that they accurately recognize input data, particularly when they are used in autonomous vehicles or for medical services. In this study, we propose a data correction method for increasing the accuracy of an unknown classifier by modifying the input data without changing the classifier. This method modifies the input data slightly so that the unknown classifier will correctly recognize the input data. It is an ensemble method that has the characteristic of transferability to an unknown classifier by generating corrected data that are correctly recognized by several classifiers that are known in advance. We tested our method using MNIST and CIFAR-10 as experimental data. The experimental results exhibit that the accuracy of the unknown classifier is a 100% correct recognition rate owing to the data correction generated by the proposed method, which minimizes data distortion to maintain the data's recognizability by humans.

• Proceedings of the KIPE Conference
• /
• 2000.07a
• /
• pp.237-240
• /
• 2000

A new control method using average comparison strategy have been proposed in this paper. This control method realizes sinusoidal input and output current. unity input displacement factor regardless of load power factor. Moreover compensation of the asymmetrical and harmonic containing input voltage is sautomatically realized and calculation time of control function is reduced.

• Smart Structures and Systems
• /
• v.8 no.2
• /
• pp.157-172
• /
• 2011

A new method for both local damage(s) identification and input excitation force identification of beam structures is presented using the dynamic response sensitivity-based finite element model updating method. The state-space approach is used to calculate both the structural dynamic responses and the responses sensitivities with respect to structural physical parameters such as elemental flexural rigidity and with respect to the force parameters as well. The sensitivities of displacement and acceleration responses with respect to structural physical parameters are calculated in time domain and compared to those by using Newmark method in the forward analysis. In the inverse analysis, both the input excitation force and the local damage are identified from only several acceleration measurements. Local damages and the input excitation force are identified in a gradient-based model updating method based on dynamic response sensitivity. Both computation simulations and the laboratory work illustrate the effectiveness and robustness of the proposed method.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.18 no.1
• /
• pp.3-10
• /
• 2008

In this paper, the optimal command input is considered in order to minimize the residual vibrations of a flexible cantilever beam when the beam simply changes its position by translation or rotation. Although a cantilever beam has many modes of vibration, it is shown that the consideration of the first mode is sufficient in this case. Thus, the problem becomes a single-degree-of-freedom system subjected to a ground excitation. Two simple methods are proposed to find the optimal command input based on the shock response spectrum (SRS). The first method is the simplest and can be applied to lightly damped cases, and the second method is applicable to more general problems. The second method gives almost the same results as the input shaping method. However the proposed method gives a easier and clearer control strategy.

• The Transactions of the Korean Institute of Power Electronics
• /
• v.26 no.2
• /
• pp.106-111
• /
• 2021

This study proposes a control method to mitigate the influence of input capacitors in photovoltaic power curtailment. The influence is analyzed by the power flow in the photovoltaic system. In conventional power curtailment, the power injected to the grid may be increased momentarily because the influence of the input capacitor on the power injected to the grid is not considered. The proposed method limits the change rate of photovoltaic array voltage to prevent a momentary increase in the power injected to the grid. The effectiveness of proposed method, which reduces power overshoot, is verified by experimental tests. The proposed method enables the power grid to operate stably in photovoltaic power curtailment.

• Journal of Institute of Control, Robotics and Systems
• /
• v.5 no.8
• /
• pp.883-888
• /
• 1999

To obtain high positioning auccuracy for a long, flex bleman Ipulator, residual vibration must be removed from the tip motion. But it is difficult to control the vibration of low frequency. There are open-loop and closed loop methods in the elimination of the residual vibration. We inroduce input shaping technique has been used as a simple open-loop method of controlling the residual vibration of a flexible manipulator. Design of input shaper in the continuous time domain is complicated. This paper presents a new technique that designs input shaper in the z-domain and analyzes input shaping method in the z-domain. This technique is simple and easy to design input shaper.

• Nuclear Engineering and Technology
• /
• v.33 no.5
• /
• pp.457-467
• /
• 2001

It is well known that the performance of a fuzzy neural network strongly depends on the input features selected for its training. In its applications to sensor signal estimation, there are a large number of input variables related with an output As the number of input variables increases, the training time of fuzzy neural networks required increases exponentially. Thus, it is essential to reduce the number of inputs to a fuzzy neural network and to select the optimum number of mutually independent inputs that are able to clearly define the input-output mapping. In this work, principal component analysis (PCA), genetic algorithms (CA) and probability theory are combined to select new important input features. A proposed feature selection method is applied to the signal estimation of the steam generator water level, the hot-leg flowrate, the pressurizer water level and the pressurizer pressure sensors in pressurized water reactors and compared with other input feature selection methods.

• Proceedings of the KIEE Conference
• /
• 2006.04a
• /
• pp.54-56
• /
• 2006

To get the robust and reliable input command, we designed shaping function for target tracking system with commander's handle. Input signals of the commander's handle are generated by human operator. It is response of the human to reduce the error between target and gun. But, tracking error while operator aim a moving target manually gives poor system performance. Input noise, particularly, affects hit accuracy as the system performance. We proposed the design method of input command shaping to reduce the Input noise and to improve the operation ability and convenience. We performed the experiments with combat vehicle, example of Target Tracking System, to show the proposed method is efficient and practical.

• Journal of Electrical Engineering and Technology
• /
• v.11 no.4
• /
• pp.915-920
• /
• 2016

A power system designer may have difficulties in choosing a suitable multiple-input converter topology for a specific target application because each multiple-input converter topology presented in the literature has its own advantages and disadvantages. In this perspective, this paper presents a systematic topology selection method for multiple-input converters with three comparison criteria including cost-saving effect, modularity potential and flexibility. Based on these criteria, this paper proposes a strategic flow chart example for choosing a proper multiple-input converter topology. This flow chart will provide a powerful selection tool to a power system designer when he or she chooses a specific multiple-input converter for a given application.

• Wind and Structures
• /
• v.17 no.4
• /
• pp.451-464
• /
• 2013

This study develops an extended inverse input estimation algorithm with intelligent adaptive fuzzy weighting to effectively estimate the unknown input wind load of nonlinear structural systems. This algorithm combines the extended Kalman filter and recursive least squares estimator with intelligent adaptive fuzzy weighting. This study investigated the unknown input wind load applied on a tower structural system. Nonlinear characteristics will exist in various structural systems. The nonlinear characteristics are particularly more obvious when applying larger input wind load. Numerical simulation cases involving different input wind load types are studied in this paper. The simulation results verify the nonlinear characteristics of the structural system. This algorithm is effective in estimating unknown input wind loads.