Regression Analysis in Interrelated Electromechanical Systems Based on the Support Vector Method
Keywords:
Regression Analysis, Target Control Function, Electromechanical SystemAbstract
Purpose: In some cases control of automated electromechanical systems implies installing regression connections between controlled
coordinates. The support vector method can be successfully used not only to solve problems of classification and diagnosis but also to
conduct a regression analysis. The goal of this paper is to substantiate an opportunity of applying the support vector method to establish
relations between controlled coordinates in an interrelated electric drive with a discrete time equalizer. Methods: There have been
studied particularities and advantages of the support vector regression, ways of using the support vector method in an interrelated
electromechanical system. Results: Mathematical description and analytical framework of the regression algorithm based on the support
vector machine have been analyzed; there has been proposed a functional diagram of an electromechanical system with interrelated
control based on discrete time equalizers where a control target function is formed by the regression model; an example of the regression
analysis programmatically using MatLab SVM Toolbox has been given. It has been stated that the reference target function which is
formed as a result of SVR-analysis can be changed by retraining a SVR algorithm and it does not require selection of special regression
dependencies. Practical relevance: The research results can be applied in design of interrelated electromechanical systems which target
function is formed by means of the regression analysis.