• 제어로봇시스템학회:학술대회논문집
• /
• 2000.10a
• /
• pp.462-462
• /
• 2000

In this paper, genetic algorithm (GA) is the technique to search for the optimal structures (i,e., the kind of neural network, the number of hidden neuron, ..) of the neural networks which are used approximating a given nonlinear function, In this paper, we used multi layer feed-forward neural network. The decision method of synapse weights of each neuron in each generation used back-propagation method. In this study, we simulated nonlinear function approximation in the temperature control system.

• Journal of Institute of Control, Robotics and Systems
• /
• v.15 no.4
• /
• pp.413-419
• /
• 2009

A decentralized adaptive control method is proposed for large-scale systems with unknown time-delayed nonlinear interconnections unmatched in control inputs. It is assumed that the time-delayed interaction terms are bounded by unknown nonlinear bounding functions. The nonlinear bounding functions and uncertain nonlinear functions of large-scale systems are compensated by the function approximation technique using neural networks. The dynamic surface control method is extended to design the proposed memoryless local controller for each subsystem of uncertain nonlinear large-scale time delay systems. Therefore, although the interconnected systems consist of a large number of subsystems, the proposed controller can be designed simply. We prove that all the signals in the total closed-loop system are semiglobally uniformly bounded and the control errors converge to an adjustable neighborhood of the origin. Finally, an example is given to demonstrate the effectiveness and applicability of the proposed scheme.

• 제어로봇시스템학회:학술대회논문집
• /
• 2004.08a
• /
• pp.407-412
• /
• 2004

This paper presents a new algorithm for the closed-loop $H_{\infty}$ composite control of singularly perturbed nonlinear systems with a exogenous disturbance, using the successive Galerkin approximation(SGA). The singularly perturbed nonlinear system is decomposed into two subsystems of a slow-time scale and a fast-time scale via singular perturbation theory, and two $H_{\infty}$ control laws are obtained to each subsystem by using the SGA method. The composite control law that consists of two $H_{\infty}$ control laws of each subsystem is designed. One of the purposes of this paper is to design the closed-loop $H_{\infty}$ composite control law for the singularly perturbed nonlinear systems via the SGA method. The other is to reduce the computational complexity when the SGA method is applied to the high order systems.

• Structural Engineering and Mechanics
• /
• v.65 no.5
• /
• pp.513-521
• /
• 2018

In this paper, the resonance response of spar-type floating platform in coupled heave and pitch motion is investigated using a CPU time-effective numerical method. A coupled nonlinear 2-DOF equation of motion is derived based on the potential wave theory and the rigid-body hydrodynamics. The transient responses are solved by the fourth-order Runge-Kutta (RK4) method and transformed to the frequency responses by the digital Fourier transform (DFT), and the first-order approximation of heave response is analytically derived. Through the numerical experiments, the theoretical derivation and the numerical formulation are verified from the comparison with the commercial software AQWA. And, the frequencies of resonance arising from the nonlinear coupling between heave and pitch motions are investigated and justified from the comparison with the analytically derived first-order approximation of heave response.

• Journal of the Korean Mathematical Society
• /
• v.51 no.2
• /
• pp.267-288
• /
• 2014

We propose and analyze two-scale product approximation for semilinear heat equations in the mixed finite element method. In order to efficiently resolve nonlinear algebraic equations resulting from the mixed method for semilinear parabolic problems, we treat the nonlinear terms using some interpolation operator and exploit a two-scale grid algorithm. With this scheme, the nonlinear problem is reduced to a linear problem on a fine scale mesh without losing overall accuracy of the final system. We derive optimal order $L^{\infty}((0, T];L^2({\Omega}))$-error estimates for the relevant variables. Numerical results are presented to support the theory developed in this paper.

• 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.

• Journal of the Korean Society for Precision Engineering
• /
• v.25 no.12
• /
• pp.89-99
• /
• 2008

In this paper, the position tracking control problem of the servo system with nonlinear dynamic friction is issued. The nonlinear dynamic friction contains a directly immeasurable friction state variable and the uncertainty caused by incomplete parameter modeling and its variations. In order to provide the efficient solution to these control problems, we propose the composite control scheme, which consists of the robust friction state observer, the FNN approximator and the approximation error estimator with sliding mode control. In first, the sliding mode controller and the robust friction state observer is designed to estimate the unknown internal state of the LuGre friction model. Next, the FNN estimator is adopted to approximate the unknown lumped friction uncertainty. Finally, the adaptive approximation error estimator is designed to compensate the approximation error of the FNN estimator. Some simulations and experiments on the servo system assembled with ball-screw and DC servo motor are presented. Results show the remarkable performance of the proposed control scheme. The robust friction state observer can successfully identify immeasurable friction state and the FNN estimator and adaptive approximation error estimator give the robustness to the proposed control scheme against the uncertainty of the friction parameters.

• Proceedings of the Korea Electromagnetic Engineering Society Conference
• /
• 2003.11a
• /
• pp.649-659
• /
• 2003

OFDM/FH communication system Is widely used in the wireless communication for the large capacity and high-speed data transmission. However, phase noise and PAPR (peak-to-average power ratio) are the serious problems causing performance impairment. In this paper, PLL (phase locked loop) frequency synthesizer with high switching speed is used for the phase noise model. SSPA and TWTA are considered for the nonlinear HPA model. Under these conditions and by approximating $e^{j{\phi}[m]}$ into $1 + j{\phi}[m]-\frac{1}{2}{\phi}^2[m]$ for the phase noise nonlinear approximation, SINR (signal-to-interference-noise-ratio) with nonlinear HPA and phase noise is derived in the OFDM/FH system. The bit error probabilities (BER) are found by computer simulation method and semi-analytical method. The simulation results closely match with the semi-analytical results.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.35C no.2
• /
• pp.56-64
• /
• 1998

Multi-layered perceptrons that are a nonlinear neural network model, have been widely used for various applications mainly thanks to good function approximation capability for nonlinear fuctions. However, for digital hardware implementation of the multi-layere perceptrons, the quantization scheme using "look-up tables (LUTs)" is commonly employed to handle nonlinear signmoid activation functions in the neworks, and thus requires large amount of storage to prevent unacceptable quantization errors. This paper is concerned with a new effective methodology for digital hardware implementation of multi-layered perceptrons, and proposes a "piecewise affine approximation" method in which input domain is divided into (small number of) sub-intervals and nonlinear sigmoid function is linearly approximated within each sub-interval. Using the proposed method, we develop an expression and an error backpropagation type learning algorithm for a multi-layered perceptron, and compare the performance with the quantization method through Monte Carlo simulations on XOR problems. Simulation results show that, in terms of learning convergece, the proposed method with a small number of sub-intervals significantly outperforms the quantization method with a very large storage requirement. We expect from these results that the proposed method can be utilized in digital system implementation to significantly reduce the storage requirement, quantization error, and learning time of the quantization method.quantization method.

• Communications for Statistical Applications and Methods
• /
• v.18 no.1
• /
• pp.131-136
• /
• 2011

The original Bates-Watts framework applies only to the complete parameter vector. Thus, guidelines developed in that framework can be misleading when the adequacy of the linear approximation is very different for different subsets. The subset curvature measures appear to be reliable indicators of the adequacy of linear approximation for an arbitrary subset of parameters in nonlinear models. Given the specific mean shift outlier model, the standard approaches to obtaining test statistics for outliers are discussed. The accuracy of outlier tests is investigated using subset curvatures.