• The Journal of the Korea institute of electronic communication sciences
• /
• v.16 no.4
• /
• pp.625-632
• /
• 2021

This paper studies the effect of the number of filters in the CNN (Convolutional Neural Network) layer on the performance of a noise attenuator. Speech is estimated from a noised speech signal using a 64-neuron, 16-kernel CNN filter and an error back-propagation algorithm. In this study, in order to verify the performance of the noise attenuator with respect to the number of filters, a program using Keras library was written and simulation was performed. As a result of simulation, it can be seen that this system has the smallest MSE (Mean Squared Error) and MAE (Mean Absolute Error) values when the number of filters is 16, and the performance is the lowest when there are 4 filters. And when there are more than 8 filters, it was shown that the MSE and MAE values do not differ significantly depending on the number of filters. From these results, it can be seen that about 8 or more filters must be used to express the characteristics of the speech signal.

• ETRI Journal
• /
• v.15 no.3
• /
• pp.1-26
• /
• 1994

Simulated annealing is an attractive, but expensive, heuristic method for approximating the solution to combinatorial optimization problems. Attempts to parallel simulated annealing, particularly on distributed memory multicomputers, are hampered by the algorithm's requirement of a globally consistent system state. In a multicomputer, maintaining the global state S involves explicit message traffic and is a critical performance bottleneck. To mitigate this bottleneck, it becomes necessary to amortize the overhead of these state updates over as many parallel state changes as possible. By using this technique, errors in the actual cost C(S) of a particular state S will be introduced into the annealing process. This paper places analytically derived bounds on this error in order to assure convergence to the correct optimal result. The resulting parallel simulated annealing algorithm dynamically changes the frequency of global updates as a function of the annealing control parameter, i.e. temperature. Implementation results on an Intel iPSC/2 are reported.

• Journal of Institute of Control, Robotics and Systems
• /
• v.12 no.4
• /
• pp.339-345
• /
• 2006

An analysis methodology of Decentralized Dynamic Surface Control (DDSC) for the large-scale interconnected nonlinear systems is presented in this paper. While the centralized DSC approach proposed in [14] has a difficulty to check the quadratic stability for the large-scale systems numerically due to dramatic increases of the order of overall augmented error dynamics, DDSC is relatively easy to check the quadratic stability since lower order error dynamics of individual subsystems are used. Then, a systematic procedure for designing DDSC will be developed. Furthermore, after a quadratic function containing a reachable set is defined, it will be calculated numerically to indicate the performance of DDSC in the framework of convex optimization. Finally an illustrative example will be given for showing the advantages of DDSC compared with other decentralized nonlinear control techniques.

• Computational Structural Engineering
• /
• v.6 no.3
• /
• pp.125-137
• /
• 1993

Recently, many researches are being dealt with the adaptive method for improving the accuracy of finite element solution. This paper deals with rh-methods that are the combination of r and h-method ; r-method is to relocate the nodes for the grid optimization, h-method is to divide the elements with great error into the equal shape. As a results, rh-method has the same error decrease and convergence as h-method in the same degree of freedom, but it has more exact result of finite element in the state of restraining degrees of freedom than h-method alone.

• 제어로봇시스템학회:학술대회논문집
• /
• 1993.10b
• /
• pp.179-184
• /
• 1993

In this paper, an optimal motion control scheme is proposed for robot manipulators. A simple explicit solution to the Hamilton-Jacobi equation is presented. The optimization of motion control is based on the mininization of the torque term affecting the kinetic energy and the augmented error which has the first-order stable dynamics for the position and velocity tracking error. In the presence of parametric uncertainty, an adaptive control scheme using the optimal principle is proposed. The global stability of the closed-loop system is guaranteed by the Lyapunov stability approach, The effectiveness and feasibility of the proposed control schemes are shown by simulation results.

• Journal of the Korean Society for Precision Engineering
• /
• v.21 no.2
• /
• pp.92-99
• /
• 2004

The errors can cause the serious loss of the performance of a precision machine system. In this paper, we proposed the method of allocating the alignment tolerances of the parts and applied this method to get the optimal tolerances of a Confocal Scanning Microscope. In general, tight tolerances are required to maintain the performance of a system, but a high cost of manufacturing and assembling is required to preserve the tight tolerances. The purpose of allocating the optimal tolerances is minimizing the cost while keeping the high performance of the system. In the optimal problem, we maximized the tolerances while maintaining the performance requirements. The Monte Carlo Method, a statistical simulation method, is used in tolerance analysis. Alignment tolerances of optical components of the confocal scanning microscope are optimized to minimize the cost and to maintain the observation performance of the microscope. We can also apply this method to the other precision machine system.

• Journal of the Korean Operations Research and Management Science Society
• /
• v.31 no.3
• /
• pp.1-10
• /
• 2006

In this paper we deal with a cell planning issue arisen in a CDMA based multicast network. In a CDMA based wireless network, a terminal can significantly reduce the bit error rate via the cohesion of data streams from multiple base stations. In this case, multiple base stations have to be operated according to a common time line. The cells whose base stations are operated as such are called soft-combined cells. Therefore, a terminal can take advantage of error rate reduction, if the terminal is in a soft-combined cell and at least one neighboring cell is also soft-combined. However, as soft-combining operation gives heavy burden to the network controller, the limited number of cells can be soft-combined. Our problem us to find a limited number of soft-combined cells such that the benefit of the soft-combining operation is maximized.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.5 no.6
• /
• pp.1089-1093
• /
• 2001

A H$\infty$ control theory is applied to the design problem of synthesis filters in a mutirate filter bank. We select a desired pure time-delay system as reference model, and then consider the error system between the mutirate filter bank and the reference model. 1'he synthesis filters minimize the ι$_2$-induced norm of the error system.

• Journal of Power System Engineering
• /
• v.4 no.3
• /
• pp.62-71
• /
• 2000

A mathematical modeling for a magnetic levitation system is proposed using the Taylor series expansion of differential function for obtaining linearity. It is confirmed that this kind of linear approximation method can be used to the modeling of a magnetic levitation system. The two-degree-of-freedom optimal servo system for a constant reference signal is proposed using the LQ optimization technique. An additional state feedback is introduced at the output of the integrator to cancel the integral action for reference signal if there is no modeling error of the plant and no disturbance input to the plant. When the modeling error or the disturbance input exists, the integral effect appears. The system has a free parameter which can b used to tune the effect of the integral compensation.

• Journal of the Korean Graphic Arts Communication Society
• /
• v.16 no.3
• /
• pp.121-131
• /
• 1998

The object color does not look same under the different light source. It depends on the surface spectral reflectance and the spectral distribution of light source. Therefore we should find the surface spectral reflectance of object color and the spectral distribution of light source for color reproduction. Using Wiener estimation, we can estimate the spectral reflectance from low dimensional images obtained with multi-band image acquisition system. The kind and the number of imaging filters have the effect on the estimation of the spectral reflectance. Therefore it is important that optimal filters are selected to minimize the error of the result. In this paper, we describe methods to select optimal filters with minimum error between measured and estimated surface spectral reflectance and to estimate surface spectral reflectance of Munsell color chart from six multi-band images by using Wiener estimation.