DocumentCode :
3655949
Title :
Method and parallelization algorithms of synthesis of empirical models taking into account the measurement errors
Author :
M.I. Gorbachuk;A.M. Lazor;M.S. Pasyeka;V.V. Bandyra;I.Yu. Yurchak
Author_Institution :
Ivano-Frankivsk national technical University of oil and gas, 76019, Karpatska str., 15, Ukraine
fYear :
2015
Firstpage :
319
Lastpage :
327
Abstract :
In this paper, the task - to investigate the effect of measurement errors on the accuracy of building empirical models. Assumed that measurement error as input and output variables are additive independent and normally distributed. Based on the assumptions made, formed a criterion approximation, which depends not only on the parameters of empirical model, but also from measurement errors. Minimizing the resulting approximation criterion allowed to obtain a system of algebraic equations, whose solutions are the unknown parameters of the model and measurement errors allowable values of process parameters. Solution of the problem is possible only in the case where the known structure of empirical models. In most cases, the researcher has no such information. For optimal synthesis model for the structure of genetic algorithm that selects the optimal model on the set of experimental data, which are called the set checkout. The method and an algorithm has been tested on experimental data, which is obtained in the operation of gas compressor units. The results allow to state that under the existing mass measurements of process parameters of gas pumping units of measurement errors significantly affect the accuracy of building empirical models.
Publisher :
ieee
Conference_Titel :
Experience of Designing and Application of CAD Systems in Microelectronics (CADSM), 2015 13th International Conference The
Type :
conf
DOI :
10.1109/CADSM.2015.7230867
Filename :
7230867
Link To Document :
بازگشت