(Хмельницький національниий університет, 2025-12) Goroshko, Andrii; Zembytska, Maryna
The article considers the problem of identifying technological processes based on experimental data using statistical methods. It is shown that in cases where the result is influenced by several independent factors, the distribution of product parameters can be multimodal. A model for describing the sample in the form of a linear mixture of normal densities is proposed, which allows to correctly reproduce the behavior of complex technological processes. Methods for estimating mixture parameters are presented, in particular, maximum likelihood, the method of moments and the method of least squares. It is shown that the use of such models increases the accuracy of forecasting and assessing the reliability of products. The results obtained can be used for product quality control and optimization of standards.