Abstract: In this paper, a new RBF neural network (RBFNN) algorithm, called ar-RBFNN, is presented. In traditional RBFNNs based on clustering algorithm, called oRBFNN in this paper, the width of the basis function-Gaussian function, or called radius, ignored the effect of numbers in different clusters, or density of data points. New algorithm considers radius is effect to performance of algorithms in problem of function approximation. Mean Square Error is used to evaluate performances of two algorithms, oRBFNN and ar-RBFNN algorithms. Several experiments in function approximation show ar-RBFNN is better than oRBFNN.
605
Authors: Xiao Lei Zhang, Jin Ming Wu
Abstract: The curve and surface fitting problem is very important in CAD and CAGD. However, it is important to construct a suitable function to interpolate or approximate which satisfies the underlying constraints since we have some additional information that is confined to interpolation or approximation. In this paper, we discuss the positive approximation for positive scattered data of any dimensionality by using radial basis functions. The approach is presented to compute positive approximation by solving a quadratic optimization problem. Numerical experiments are provided to illustrate the proposed algorithm is flexible.
683
Authors: Igor Grešovnik, Tadej Kodelja, Robert Vertnik, Božidar Šarler
Abstract: A framework for optimization of process parameters in material processing and production is described. The framework is designed for effective set up and solution of optimization problems as part of process design, as well as to support development of numerical models by inverse identification of model parameters. The general framework is outlined, which has been supplemented by a neural networks module in order to enable real time decision support. Simulator based on meshless method with radial basis functions (RBF) has been utilized.
838
Authors: Piotr Czop, Damian Slawik, Grzegorz Wszołek, Dawid Jakubowski, Antoni Skrobol
Abstract: This paper proposes an analytical tool that supports the design process of a disc spring valve system used in car dampers. The proposed analytical tool obtains a key design characteristic of a valve, which is the flow rate and the corresponding maximum stress level in the stack of plates, as a function of a pressure load. The tool is prepared based on the cases produced by a first-principle model using a finite element approach. The finite element model was calibrated based on experimental results to provide accurate results in the entire range of input parameters.
1365
Authors: Shou Jun Li, Xiao Ping Ma, Hong Yu
Abstract: It is an important means of hydrological data analysis for drawing hydrological data curve. The paper conducts a study on drawing method of stage-discharge curve in two aspects including BP neural network approximation and curve fitting, according to data extracted from a hydrologic station located in Suqian section of Beijing-Hangzhou Canal. Normalization of the input sample is processed in order to caculate conveniently and prevent partial neurons to supersaturate. Then, neuronal number is determined by method of heuristics. And the transfer function and training function are finalized on the premise of target error 0.0001.Error analysis is performed after simulation of BP network approximation. 2- and 3-order curve fitting is done based on principle of least squares of polynomial fitting, then followed by error analysis. Comparison of both methods comes to the conclusion that approximation of BP network for a given data is more accurate than that of curve fitting.
4115