Adaptive piecewise linear approximation of difficult-to-compute functions

Journal Title: Проблеми машинобудування - Year 2018, Vol 21, Issue 2

Abstract

The solution of many theoretical and applied problems requires that some functional dependencies be substituted into others, which are more convenient for the implementation of a specific mathematical problem. At the same time, information about the character of the original function can be insufficient, and the function itself can be considered to be difficult to compute. The accuracy of such an approximation depends on the methods used, the character of the original function, as well as the number and choice of grid points. The easiest way of building such an approximation is doing it on a uniform grid of points, which does not always provide an acceptable result. The purpose of this paper is to develop effective adaptive methods of approximating functions for the problems aimed at searching for the lengths of curves and calculating integrals under conditions of limited information about the character of the original function and the presence of its derivatives. An adaptive approach to the approximation of a wide class of one-dimensional functions is proposed in the paper. For this approximation a piecewise linear approximation with a simple mechanism of exponential adaptive feedback step process control is used. The possibilities of this approach are considered, using the problems of calculating the lengths of curves and values of definite integrals. The specifics of the application of the suggested approach are detailed for each case. The approach does not require an initial allocation of grid points. The method ensures the required accuracy in automatic mode. The result is realized in a single pass without any preliminary transformations. The reliability of the obtained results is confirmed by solving the known test examples. The results of calculating a number of definite integrals with different nature of integrand are presented. The calculation results by the proposed method are compared with the data obtained by the usual trapezoid method. A high efficiency of the proposed approach is established. The proposed method opens the way for creating effective means for solving numerical integration and differentiation problems, as well as integral and differential equations and so on.

Authors and Affiliations

G. A. Sheludko, S. V. Ugrimov

Keywords

Related Articles

Experimental Strength Analysis of Variable Stiffness Waffel-Grid Cylindrical Compartments. Part 1. Experimental Procedure

This paper proposes a method of the experimental study of the static stress-strain state (SSS) of the variable stiffness tail section of a launch vehicle (LV). The tail section consists of a body and a support ring. The...

Experimental Study of the Effects of a Constant and Variable Electric Fields on the Strength Properties of the Blade Steel Surface Layer

It is known that in wet steam turbines the wet steam flow is electrified when passing through the turbine flow path. In this, the mutual electrification of the condensed moisture flow and flow path surfaces occurs. Until...

Modeling Crack Initiation in a Composite Under Bending

It is known that multi-component structures are more reliable and durable than homogeneous ones. At the design stage of new structures from composite materials, it is necessary to take into account the cases when cracks...

Peculiarities of Identifying a Power-generating Single-shaft GTE Mathematical Model

The processes of designing and developing gas turbine engines (GTE) are based on using mathematical models (MM), reflecting the physical picture of engine operation processes. One of the ways of improving the MM validity...

Substantiation of Boundary Accelerations of Roller Forming Unit Optimal Reversal Mode According to Fourth-Order Acceleration

In order to increase the reliability and durability of a roller forming unit, we calculated a combined mode of the reciprocating movement of a forming trolley with the reversal according to the fourth-order acceleration...

Download PDF file
  • EP ID EP622078
  • DOI 10.15407/pmach2018.02.060
  • Views 55
  • Downloads 0

How To Cite

G. A. Sheludko, S. V. Ugrimov (2018). Adaptive piecewise linear approximation of difficult-to-compute functions. Проблеми машинобудування, 21(2), 60-67. https://europub.co.uk/articles/-A-622078