Applying covariance analysis to determine the factor influence on the control parameter in colorimetric study

Abstract

The paper considers solutions to a scientific and practical problem of applying covariance analysis to determine the factor influence on the functional transformation of the control parameter in colorimetric study. The study implies determining the factor influence on the additive and multiplicative components of the measurement error of the colorimetric control parameter to assess the validity of conclusions about the factor influence on the transformation of the control parameter. The limitations on the number of basic levels (control parameters) and the factors influencing the result of colorimetric control are determined. During the study, the equations to assess the validity of statistical conclusions about the informational significance of the colorimetric control indicators for a simplified model of cross-classification were obtained.
 The need for the study is due to the fact that in colorimetric control of cereal grains, the measurement uncertainty of the results of measuring the values of the controlled indicators at given levels of the control parameter is quite high.
 In modern industry, colorimetry has a number of advantages over other methods, such as weight analysis. Colorimetric determinations are performed much faster. In the case of weight analysis, the chemical reaction is at the beginning of the determination followed by a series of long operations, while in the case of colorimetry, the colors are compared immediately after the chemical reaction. Therefore, the colorimetric method belongs to the methods of express control.
 The main task of colorimetric express control (in our case) is to determine the quality of cereal grains. The task is important and relevant because grain, namely flour, is the main ingredient of many foods, e.g. bread, pasta, pastries, and cookies. Quality products can be obtained only from quality raw materials. Thus, at present time, it is relevant to determine the quality of cereal grains, and flour as the derived product, in Ukraine.

Dispersion analysis in colorimetric control

“…At the same time, a researcher may be interested in how significant the influence of one or another factor on the dispersion of the results of observations of the initial value is. To study the influence of factors that interfere with the initial value (feedback), their overall estimation, ranking and selecting the most significant ones among them, all methods of screening quantitative factors and regression analysis methods are obviously unsuitable since they involve measuring the levels of the factors being studied [7,8].…”

“…To solve this kind of problems, researchers use different techniques and ways of organizing sample data, the essence of which consists in applying different plans for classifying observations according to possible sources of dispersion. The processing of such classified data is carried out using the methods of dispersion analysis, which allow decomposing general dispersion into components caused by the influence of the levels of factors of interest to the researcher [9].…”

“…The model of the influence of factors on the measurement results of a single control indicator K (grain moisture) is characterized by two influencing factors (grain temperature, illumination). It has the following form [9]:…”

“…h abcv is a random remainder. The simplified model ( 1) can be reduced to two two-factor cross-classification models [9]:…”

“…For this, the number of groups of the results of observing the values of the control indicator K should be the same for all models. This corresponds to the same number of terms in the source data table [9]. At the same time, the number of values of the control indicator K in each of the groups should be the same g. For this, the number of groups of the results of observing the values of the control indicator K should be the same for all models.…”