A Method for Periodical Phenomena Analysis

Abstract

This paper is proposing a statistical method, useful for analyzing periodical phenomena whose equations are impossible to be solved analytically. The input data consist in a collection of tabled functions, (numerically determined), named further database. Least square method based algorithms, presented in the first part of the paper, are applied to this database. First, a trigonometric regression algorithm will find the approximating Fourier coefficients. Finally a multiple regression algorithm to fit a polynomial type function is introduced. Its input data are the Fourier coefficients, the Fourier analysis results, or whatever collection of experimental data corresponding to a set of variables. The final part of the paper is dedicated to an example, illustrative for these.