76-84

UDC 621.3.049.77:519.242
DOI: 10.15350/2306-2819.2018.3.76

Development of Experiment Designs In Case of the Linear Drift of the Quality index of an Electronic Engineering Product

Yu. V. Zaharov¹, N. G. Moiseev¹, A. D. Kainova²
¹Volga State University of Technology,
3, Lenin Square, Yoshkar-Ola, 424000, Russian Federation
E-mail: ZaharovYV@volgatech.net
²Russian State Social University,
Building 1, 4, Wilhelm Pieck Street, Moscow, 129226, Russian Federation
E-mail: info@rqsu.net

ABSTRACT

Introduction. When constructing a mathematical model of the dependence of electronic engineering product quality index on factors, determining its numerical value, there are a number of problems, concerning experiment implementation under production conditions. The main problem is a time drift of the product quality index due to variables, changing in time, and their control is not possible. These are diurnal and seasonal variations of environmental conditions; properties of raw materials, materials and half-finished products; manufacturing equipment deterioration; diffusant exhaustion; variation of catalyst characteristics and others. The purpose of the work is the development of designs of a complete factorial experiment (CFE) for adequate mathematical modeling in case of the linear continuous time drift of the product quality index under production conditions. Results. During the standard experimental procedure of CFE design due to the time drift, measured values of the quality index of the electrical engineering product do not reflect the object response to planned (according to the experiment design) combination of levels of input factors and the response distortion that is their “displacement” from real values occurs. The use of “biased” data leads to an inadequate mathematical model, false conclusions, concerning the influence of various input factors on the product quality index that makes it difficult to estimate the manufacturing process state. It is shown that if the standard (existing) experimental procedure for the CFE design is changed, then the adequate mathematical model can be obtained under conditions of the time drift of the product quality index. The efficient designs of CFE were developed, when the number of input factors was k = 2-5. For k = 2 and k = 3 experiment designs were obtained by means of complete enumeration 4! and 8! of possible experimental procedures. In order to obtain experiment designs for k = 4 and k = 5 by means of complete enumeration, the consideration of 16! and 32! of possible experimental procedures is required, which is complicated. The heuristic algorithm for incomplete enumeration of the set of possible experimental procedures was developed. The samples of developed experiment designs were presented. Conclusion. The modified designs of the complete factorial experiment of the type (k = 2-5), which allow obtaining the adequate mathematical model in case of the linear continuous time drift of the quality index of the electronic engineering product were developed.

KEYWORDS

mathematical model; complete factorial experiment; linear time drift; unbiased estimates of model coefficients.

FULL TEXT (pdf)

REFERENCES

1. Sidnyaev N. I. Teoriya planirovanii ehksperimenta i analiz statisticheskih danny [The theory of planning the experiment and analysis of statistical data]. Moscow: Izd-vo Yurajt, 2014. 495 p.
2. Zaharov Yu. V., Moiseev N. G. Vybor pokazatelej pri ocenke kachestva izdelij ehlektronnoj tekhniki [Selection of indicators when evaluating electronic engineering item quality]. Vestnik Povolzhskogo gosudarstvennogo tekhnologicheskogo univer-
siteta. Ser.: Radiotekhnicheskie i infokommunikatsionnye sistemy [Vestnik of Volga State University of Technology. Ser.: Radio Engineering and Infocommunication Systems]. 2015. No 4 (28). Pp. 75-84.
3. Meledina T. V., Danina M. M. Metody planirovaniya i obrabotki rezul'tatov nauchnyh issledovanij [Methods of planning and processing research results]. SPb .: NRU ITMO, 2015. 110 p.
4. Zazhigaev L.S., Kishyan A.A., Romanikov Yu.I. Metody planirovaniya i obrabotki rezul'tatov fizicheskogo ehksperimenta [Methods of planning and processing the results of a physical experiment]. Moscow: Atomizdat, 2013. 232 p.
5. Moiseev N. G., Zakharov Yu. V. Teoriya planirovaniya i obrabotki ehksperimenta [Theory of planning and processing the experiment]. Yoshkar-Ola: Perm State Technical University, 2018. 124 p.
6. Radchenko S. G. Sistemnoe optimal'noe planirovanie regressionnogo ehksperimenta [System optimal planning of a regression experiment]. Zavodskaya laboratoriya. Diagnostika materialov [Factory Laboratory. Diagnostics of materials]. 2012. Vol. 78, No. 7. Pp. 71-75.
7. Radchenko S. G. Metodologiya regressionnogo analiza: monografiya [Methodology of regression analysis: monograph]. Kiev: “Kornichuk”, 2011. 376 p.
8. Orlov A.I. Mnogoobrazie modelej regressionnogo analiz [The Variety of Regression Analysis Models]. Zavodskaya laboratoriya. Diagnostika materialov [Factory Laboratory. Diagnostics of materials]. 2018. Vol. 84, No. 3. Pp. 68-72.

For citation: Zaharov Yu. V., Moiseev N. G., Kainova A. D. Development of Experiment Designs in Case of the Linear Drift of the Quality Index of an Electronic Engineering Product. Vestnik of Volga State University of Technology. Ser.: Radio Engineering and Infocommunication Systems. 2018. No 3 (39). Pp. 76-84. DOI: 10.15350/2306-2819.2018.3.76