
Main / Gulnara Rivalevna Kadyrova
Author: "Gulnara Rivalevna Kadyrova "
Gulnara Rivalevna Kadyrova, Ulyanovsk State Technical University, Candidate of Engineering; graduated from the Faculty of Radioengineering of Ulyanovsk Polytechnic Institute; Associate Professor at the Department of Applied Mathematics and Informatics of Ulyanovsk State Technical University; an author of monographs, textbooks, and articles in the field of statistical modeling, software information systems. [email: gulya@ulstu.ru]G. Kadyrova


Modification of the Stepbystep Regression Method for Mathematical Models of the Object Behavior Prediction
The temperature regime significantly affects the durability of the computer. Ensuring stability of the computer functioning considers the stability of the heating temperature of the main elements that should not exceed the specified values. The article discusses issues related to the timely warning about a possible violation of the temperature regime stability. In order to diagnose stability, the multivariate statistical control methods are proposed to use. Evaluation of stability is carried out with the use of two criteria: stability of the temperature average level and their dispersion. Independent parameters can be controlled with the use of standard shewhart charts. The algorithms on the basis of Hotelling statistics (for assessing stability of the middlelevel temperature measurement process) and generalized variance (for evaluation of the dispersion process stability) are used for correlated parameters. The efficiency of these algorithms can be enhanced through the analysis of nonrandom structures on the control charts, the use of the warning border, as well as application of modifications based on the cumulative amounts or moving averages weighted exponentially. The multivariate statistical technique of the computer temperature regime control including monitoring in the context of a welldeveloped process for a training sample in order to separate the parameters controlled for a group of independent and correlated ones, process analysis for assessment of control characteristics and continuous monitoring of the process with the construction of Hotelling charts and generalized dispersion with identifying possible violations of the process based on the presence of significant structures and the use of a warning border. This technique is illustrated by the example of five computer temperature regime parameters control. .



Sections: Mathematical modeling
Subjects: Mathematical modeling, Operational research. 
Kadyrova Gulnara Rivalevna, Ulyanovsk State Technical University, Candidate of Engineering; graduated from the Faculty of Radioengineering of Ulyanovsk Polytechnic Institute; Associate Professor of the Department of Applied Mathematics and Informatics of Ulyanovsk State Technical University; an author of monographs, textbooks, and articles in the field of statistical modeling, software information systems. [email: gulya@ulstu.ru] G.R. Kadyrova


Evaluation and Prediction of Technical Object Condition With the Use of Regression Models
The article presents information about the System of Optimum Regression Search (SORS) statistical package realizing the approach of adaptive regression modelling and providing an evaluation of the observation model adequacy and search of its optimal structure. The methodology means reduction of the model dimension and increase of determination accuracy of its parameters and the forecast. Efficiency of the methodology is directly proportional to dimension, the degree of noisiness, and multicollinearity of the initial data. Consequently, it allows considering its application for the description of technical objects conditions as perspective mathematical approach. One of the tasks in data analysis is the problem of choice of a comparison measure for the competitive models. The use of internal quality measures for smaller dimension models intended for forecasting cannot always give as authentic view of a particular competitive structure that is preferable as possible. The article investigates properties of the cross validation measure that is based on the complete sample of data and uses it as control data sample for familiar internal and external measures. The prospect of its application for identification of the forecasting optimal model within SORS is demonstrated in the article. Regression modelling, forecasting, methods of structural identification, quality measure, sors, statistical package.



Sections: Mathematical modeling
Subjects: Mathematical modeling, Information systems. 
