• International Journal of Control, Automation, and Systems
• /
• v.6 no.1
• /
• pp.24-34
• /
• 2008

A novel neural network model, termed the standard neural network model (SNNM), similar to the nominal model in linear robust control theory, is suggested to facilitate the synthesis of controllers for delayed (or non-delayed) nonlinear systems composed of neural networks. The model is composed of a linear dynamic system and a bounded static delayed (or non-delayed) nonlinear operator. Based on the global asymptotic stability analysis of SNNMs, Static state-feedback controller and dynamic output feedback controller are designed for the SNNMs to stabilize the closed-loop systems, respectively. The control design equations are shown to be a set of linear matrix inequalities (LMIs) which can be easily solved by various convex optimization algorithms to determine the control signals. Most neural-network-based nonlinear systems with time delays or without time delays can be transformed into the SNNMs for controller synthesis in a unified way. Two application examples are given where the SNNMs are employed to synthesize the feedback stabilizing controllers for an SISO nonlinear system modeled by the neural network, and for a chaotic neural network, respectively. Through these examples, it is demonstrated that the SNNM not only makes controller synthesis of neural-network-based systems much easier, but also provides a new approach to the synthesis of the controllers for the other type of nonlinear systems.

• Journal of Information Processing Systems
• /
• v.15 no.5
• /
• pp.1201-1210
• /
• 2019

Stock price is characterized as being mutable, non-linear and stochastic. These key characteristics are known to have a direct influence on the stock markets globally. Given that the stock price data often contain both linear and non-linear patterns, no single model can be adequate in modelling and predicting time series data. The autoregressive integrated moving average (ARIMA) model cannot deal with non-linear relationships, however, it provides an accurate and effective way to process autocorrelation and non-stationary data in time series forecasting. On the other hand, the neural network provides an effective prediction of non-linear sequences. As a result, in this study, we used a hybrid ARIMA and neural network model to forecast the monthly closing price of the Shanghai composite index and Shenzhen component index.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.13 no.2
• /
• pp.686-710
• /
• 2019

This work proposes a model to solve the problem of Network Coding over a one-session multicast network. The model is based on a system of restrictions that defines the packet flows received in the sink nodes as functions of the outgoing flows from the source node. A multicast network graph is used to derive a directed labeled line graph (DLLG). The successive powers of the DLLG adjacency matrix to the convergence in the null matrix permits the construction of the jump matrix Source-Sinks. In its reduced form, this shows the dependency of the incoming flows in the sink nodes as a function of the outgoing flows in the source node. The emerging packets for each outgoing link from the source node are marked with a tag that is a linear combination of variables that corresponds to powers of two. Restrictions are built based on the dependence of the outgoing and incoming flows and the packet tags as variables. The linear independence of the incoming flows to the sink nodes is mandatory. The method is novel because the solution is independent of the Galois field size where the packet contents are defined.

• 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.

• 제어로봇시스템학회:학술대회논문집
• /
• 1994.10a
• /
• pp.731-736
• /
• 1994

Experiment on a lab-scale pH process is carried out to evaluate the control performance of the neural linearizing control scheme(NLCS) using a radial basis function(RBF) network which was previously proposed by Kim and Park. NLCS was developed to overcome the difficulties of the conventional neural controllers which occur when they are applied to chemical processes. Since NLCS is applicable for the processes which are already controlled by a linear controller and of which the past operating data are enough, we first control the pH process with PI controller. Using the operating data with PI controller, the linear reference model is determined by optimization. Then, a IMC controller replaces the PI controller as a feedback controller. NLCS consists of the IMC controller and a RBF network. After the learning of the neural network is fully achieved, the dynamics of the process combined with the neural network becomes linear and close to that of the linear reference model and the control performance of the linear control improves. During the training, NLCS maintains the stability and the control performance of the closed loop system. Experimental results show that the NLCS performs better than PI controller and IMC for both the servo and the regulator problems.

• The Transactions of The Korean Institute of Electrical Engineers
• /
• v.59 no.7
• /
• pp.1320-1326
• /
• 2010

In this paper, we develop a design methodology of granular-based neurocomputing networks realized with the aid of the clustering techniques. The objective of this paper is modeling and evaluation of approximation and generalization capability of the Linear-Type Superconducting Power Supply (LTSPS). In contrast with the plethora of existing approaches, here we promote a development strategy in which a topology of the network is predominantly based upon a collection of information granules formed on a basis of available experimental data. The underlying design tool guiding the development of the granular-based neurocomputing networks revolves around the Fuzzy C-Means (FCM) clustering method and the Radial Basis Function (RBF) neural network. In contrast to "standard" Radial Basis Function neural networks, the output neuron of the network exhibits a certain functional nature as its connections are realized as local linear whose location is determined by the membership values of the input space with the aid of FCM clustering. To modeling and evaluation of performance of the linear-type superconducting power supply using the proposed network, we describe a detailed characteristic of the proposed model using a well-known NASA software project data.

• Transactions of the Korean hydrogen and new energy society
• /
• v.26 no.3
• /
• pp.294-299
• /
• 2015

Free piston linear engine (FPLE) is a promising concept being explored in the mid-20th century. On the other hand, Arficial neural networks (ANNs) are non-linear computer algorithms and can model the behavior of complicated non-linear processes. Some researchers already studied this method to predict internal combustion engine characteristics. However, no investigation to predict the performance of a FPLE using ANN approach appears to have been published in the literature to date. In this study, the ability of an artificial neural network model, using a back propagation learning algorithm has been used to predict the in-cylinder pressure, frequency, maximum stroke length of a free piston linear engine. It is advised that, well-trained neural network models can provide fast and consistent results, making it an easy-to-use tool in preliminary studies for such thermal engineering problems.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.1624-1627
• /
• 2005

We proposed a design technique for auto-landing guidance and control system. This technique utilizes linear controller and neural network. Main features of this technique is to use conventional linear controller and compensate for the error coming from the model uncertainties and/or reference model mismatch. In this study, the multi-perceptron neural network with single hidden layer is adopted to compensate for the errors. Glide-slope capture logic for auto-landing guidance and control system is designed in this technique. From the simulation results, it is observed that the responses of velocity and pitch angle to commands are fairly good, which are directly related to control inputs of throttle and elevator, respectively.

• Journal of the Institute of Convergence Signal Processing
• /
• v.3 no.1
• /
• pp.80-86
• /
• 2002

A neural network structure which is able to perform the operations of analog addition and linear equation is proposed. The network employs Hopfkeld's model of a neuron with the connection elements specified on the basis of an analysis of the energy function. The analog addition network and linear equation network are designed by using Hopfield's A/D converter and linear programming respectively. Simulation using Pspice has shown convergence predominently to the correct global minima.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 1996.10a
• /
• pp.227-233
• /
• 1996

The Inverted Pendulum has been one of most popular nonlinear dynamic systems for the exploration of control techniques. This paper presents a new linear optimal control techniques and nonlinear neural network learning methods. The multiayered neural networks are used to add nonlinear effects on the linear optimal regulator(LQR). The new regulator can compensate nonlinear system uncertainties that are not considered in the LQR design, and can tolerated a wider range of uncertainties than the LQR alone. The new regulator has two neural networks for modeling and control. The neural network for modeling is used to obtain a more accurate model than the given mathematical equations. The neural network for control is used to overcome deficiencies by adding corrections to the linear coefficients of the LQR and by adding nonlinear effects on the LQR. Computer simulations are performed to show the applicability and a more robust regulator than the LQR alone.