Pyhteeva I. Approximation of flat discretely presented curve lines on the basis of the least squares discrete method

Українська версія

Thesis for the degree of Candidate of Sciences (CSc)

State registration number

0405U000331

Applicant for

Specialization

  • 05.01.01 - Прикладна геометрія, інженерна графіка

19-01-2005

Specialized Academic Board

К 18.819.02

Essay

New method of discrete approximation of dis-cretely presented curves (DPC) - the least squares discrete method (LSDM). Flat DPC approximation algorithm is represented by proposed method, the distin-guishing feature of which is discrete presenta-tion of the approximating function, determined by some control ordinates that influence the ordinates of the points. The dual task LSDM is solved by trans-ferring to the sphere of parameters. Suspended LSDM is considered as the basis for discrete solution of the least modulus method task. The problem of oscillation preventing under dis-crete approximation is solved by the proposed method on the basis of square programming by Langrage factor method with the usage of dis-crete function presentation. Thus, extreme solu-tion without regard for approximating function is got. The way of straightline segments of so-lution correction is proposed to enable the ap-proximation of dynamic outlines. Application of method for mathematical treatment of the experiment outcomes,and also for solution of economic-mathematical stimulation tasks, in particular, in stimulation of producing func-tions and development of producing processes prediction models in agroindustrial complex are considered. The results of researches are introduced into production at the open-end company "Mo-tor-Sich" in the chief designer department, at the open-end company "KB Berdyansksel-mash" at the chief designer department on the mathematical treatment of experiment out-comes, at Melitopol regional administration when constucting prediction models of agricul-tural crops productivity, the results of re-searches are included into educational process.

Files

Similar theses