• Journal of Electrical Engineering and Technology
• /
• v.4 no.2
• /
• pp.229-233
• /
• 2009

This paper proposes a correction method for the error inherently created by time-step approximation in finite element analysis (FEA). For a simple RL and RLC linear circuit, the error in time-step analysis is analytically investigated, and a correction method is proposed for a non-linear system as well as a linear one. Then, for a practical inductor model, linear and non-linear time-step analyses are performed and the calculation results are corrected by the proposed methods. The accuracy of the corrected results is confirmed by comparing the electric input and output powers.

• Journal of information and communication convergence engineering
• /
• v.2 no.3
• /
• pp.187-192
• /
• 2004

A method using two dimension Haar functions approximation for solving the problem of a partial differential equation and estimating the parameters of a non-linear distributed parameter system (DPS) is presented. The applications of orthogonal functions, including Haar functions, and their transforms have been given much attention in system control and communication engineering field since 1970's. The Haar functions set forms a complete set of orthogonal rectangular functions similar in several respects to the Walsh functions. The algorithm adopted in this paper is that of estimating the parameters of non-linear DPS by converting and transforming a partial differential equation into a simple algebraic equation. Two dimension Haar functions approximation method is introduced newly to represent and solve a partial differential equation. The proposed method is supported by numerical examples for demonstration the fast, convenient capabilities of the method.

• Proceedings of the Korean Society of Machine Tool Engineers Conference
• /
• 1999.05a
• /
• pp.141-146
• /
• 1999

A new method, so called MMA(Method of Moving Asymptotes) was applied to the optimization problems of non-linear functions and non-linear structures. In each step of the iterative process, tile MMA generates a strictly convex approximation subproblems and solves them by using the dual problems. The generation of these subproblems is controlled by so called 'moving asymptotes', which may both make no oscillation and speed up tile convergence rate of optimization process. By contrast in generalized dual function, the generated function by MMA is always explicit type. Both the objective and behaviour constraints which were approximated are optimized by dual function. As the results of some examples, it was found that this method is very effective to obtain the global solution for problems with many local solutions. Also it was found that MMA is a very effective approximate method using the original function and its 1st derivatives.

• Proceedings of the IEEK Conference
• /
• 2002.06c
• /
• pp.17-20
• /
• 2002

This paper proposes a new 75 fuzzy model approximation method which reduces error in nonlinear fuzzy model approximation over the V-type decision rules. Employing rational Bezier curves used in computer graphics to represent curves or surfaces, the proposed method approximates the decision rule by constructing a tractable linear equation in the highly non-linear fuzzy rule interval. This algorithm is applied to the self-adjusting air cushion for spinal cord injury patients to automatically distribute the patient's weight evenly and balanced to prevent decubitus. The simulation results indicate that the performance of the proposed method is bettor than that of the conventional TS Fuzzy model in terms of error and stability.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.20 no.12
• /
• pp.1333-1339
• /
• 2009

A MWR(Missile Warning Radar) of GVES(Ground Vehicle Equipment System) has to effectively decide the threat for a detected target. Linear Approximation Fitting(LAF) and Weighted Linear Approximation Fitting(WLAF) algorithm is proposed as algorithm for a threat decision method. The target is classified into a threat or non-threat using a boundary condition of the angular rate, and the boundary condition is determined using probability model simulation. This paper confirms the performance of proposed threat decision algorithm using measurement.

• ETRI Journal
• /
• v.15 no.2
• /
• pp.35-51
• /
• 1993

This paper presents function approximation based on nonparametric estimation. As an estimation model of function approximation, a three layered network composed of input, hidden and output layers is considered. The input and output layers have linear activation units while the hidden layer has nonlinear activation units or kernel functions which have the characteristics of bounds and locality. Using this type of network, a many-to-one function is synthesized over the domain of the input space by a number of kernel functions. In this network, we have to estimate the necessary number of kernel functions as well as the parameters associated with kernel functions. For this purpose, a new method of parameter estimation in which linear learning rule is applied between hidden and output layers while nonlinear (piecewise-linear) learning rule is applied between input and hidden layers, is considered. The linear learning rule updates the output weights between hidden and output layers based on the Linear Minimization of Mean Square Error (LMMSE) sense in the space of kernel functions while the nonlinear learning rule updates the parameters of kernel functions based on the gradient of the actual output of network with respect to the parameters (especially, the shape) of kernel functions. This approach of parameter adaptation provides near optimal values of the parameters associated with kernel functions in the sense of minimizing mean square error. As a result, the suggested nonparametric estimation provides an efficient way of function approximation from the view point of the number of kernel functions as well as learning speed.

• Structural Engineering and Mechanics
• /
• v.45 no.5
• /
• pp.613-629
• /
• 2013

In this article we will present a method to directly evaluate the critical point of a non-linear system by using the solution of a polynomial eigenvalue approximation as a starting point for an iterative non-linear system solver. This method will be used to evaluate out-of-plane buckling properties of truss structures for which the lateral displacement of the upper chord has been prevented. The aim is to assess for a number of example structures whether or not the linearized eigenvalue solution gives a relevant starting point for an iterative non-linear system solver in order to find the minimum positive critical load.

• Journal of the Korean Society of Manufacturing Process Engineers
• /
• v.19 no.6
• /
• pp.23-28
• /
• 2020

A non-linear vibration isolation system composed of a non-linear spring and a linear damper was presented in a previous study. The advantage of the proposed isolator is the simple structure of the system. When the base of the isolator is harmonically excited, the response component of the mass at the excitation frequency was approximated using three different methods: linear approximation, harmonic balance, and higher-order frequency response functions (FRFs). The method using higher-order FRFs produces significantly more accurate results compared with the other methods. The error between the exact and approximate responses does not increase monotonously with the excitation amplitude and is less than 2%.

• The Journal of the Korea Contents Association
• /
• v.21 no.3
• /
• pp.616-625
• /
• 2021

Deep neural networks are an approximation method that approximates an arbitrary function to a linear model and then repeats additional approximation using a nonlinear active function. In this process, the method of evaluating the performance of approximation uses the loss function. Existing in-depth learning methods implement approximation that takes into account loss functions in the linear approximation process, but non-linear approximation phases that use active functions use non-linear transformation that is not related to reduction of loss functions of loss. This study proposes parametric activation functions that introduce scale parameters that can change the scale of activation functions and location parameters that can change the location of activation functions. By introducing parametric activation functions based on scale and location parameters, the performance of nonlinear approximation using activation functions can be improved. The scale and location parameters in each hidden layer can improve the performance of the deep neural network by determining parameters that minimize the loss function value through the learning process using the primary differential coefficient of the loss function for the parameters in the backpropagation. Through MNIST classification problems and XOR problems, parametric activation functions have been found to have superior performance over existing activation functions.

• 전기의세계
• /
• v.22 no.1
• /
• pp.28-34
• /
• 1973

This paper treats with the sub-optimal control problem of noninear systems by approximation method. This method involves the approximation by linearization which provides the sub-optimal solution of non-linear control problems. The result of this work shows that, in the problem in which the controlled plant is characterized by an ordinary differential equation of first order, the solution obtained by this method coincides with the exact solution of problem. In of case of the second or higher order systems, it is proved analytically that this method of linearization produces the sub-optimal solution of the given problem. It is also shown that the sub-optimality of solution by the method can be evaluated by introducing the upper and lower bounded performance indices. Discussion is made on the procedure with some illustrative examples whose performance indices are given in the quadratic forms.