The prediction of a single observable time series has been achieved with varying degrees of success. The quality and duration of the prediction is dependent on many factors, the two most important being the reconstruction technique and the quantity of data. The goal of this work is to reduce the computational effort required to achieve satisfactory predictions. Without new methods, which are beyond the scope of this work, this requires a reduction in the size of the data set.

This thesis expands on earlier works using the delay vector space method and the autocorrelation function for reconstruction and applies this analysis technique to a well known non-linear dynamic system. The embedding delay and the sampling rate were varied while keeping the number of points the same in order to study the effects of varying the sampling rate. The results of this experimentation show the importance of the sampling rate and duration of the sample in the reconstruction and prediction. It is shown that the sampling duration may be more important than the number of points. It is apparent from this characteristic that a time series sampled over a longer duration may contain more information in fewer points.

Library of Congress Subject Headings

System analysis; Nonlinear theories; Chaotic behavior in systems; Dynamics

Publication Date


Document Type


Student Type


Degree Name

Mechanical Engineering (MS)

Department, Program, or Center

Mechanical Engineering (KGCOE)


Josef S. Török

Advisor/Committee Member

Hany Ghoneim

Advisor/Committee Member

Teresa Wallace


Physical copy available from RIT's Wallace Library at QA402 .R62 1994


RIT – Main Campus