RECURRENT ESTIMATION OF PARAMETERS BEING MULTIDIMENSIONAL IN TERMS
OF INPUT AND OUTPUT OF VARIOUS ORDERS OF LINEAR DYNAMIC SYSTEMS IN THE PRESENCE
OF AUTOCORRELATED NOISES IN INPUT AND OUTPUT SIGNALS 

Sandler Il'ya L'vovich, Postgraduate student, Samara State Transport University (18 1-y Bezymyanniy lane, Samara, Russia), viruskvam@bk.ru

519.254

10.21685/2072-3040-2016-4-2

Background. The use of methods to identify parameters of both linear and nonlinear objects is very widespread in various areas of science and technology. Such methods enable to increase operability of the existing automatic control systems, however not all methods allow to determine parameters of the specified models with appropriate properties (solvency and others). One of frequently applied models is a discrete linear dynamic model, multidimensional in terms of input and output of various orderz (forecasting of specific fuel consumption, electromagnetic parameters of engines and others). The purpose of the work is to develop a new mathematical method of estimation of parameters being multidimensional in terms of input and output of various orders of a discrete linear dynamic system in conditions of aprioristic uncertainty (absence of knowledge of the law of distribution of supervision hindrances) on the basis of the method of stochastic approximation, and also to create and test the software realizing the specified algorithm.
Materials and methods. The article offers a recurrent estimation algorithm for parameters that are multidimensional in terms of input and output of various orders of a discrete linear dynamic system with the non-stationary autocorrelated noises of supervision in input and output signals in the absence of knowledge of laws of distribution.
Results and conclusions. The work resulted in the development of the recurrent algorithm of estimation of parameters that are multidimensional in terms of input and output of various orders of a linear dynamic system in the presence of nonstationary autocorrelated interferences of supervision in input and output signals. The author proved strong solvency of the received estimates, as well as developed the software which will allow to increase efficiency of control systems of technological processes in various technical areas on the basis of the received algorithm.

recurrent algorithm, linear dynamic system, multidimensional system, different order, stochastic approximation, autocorrelation, aprioristic uncertainty

1. Sandler I. L. Sistemy upravleniya i informatsionnye tekhnologii [Control systems and information technologies]. 2012, no. 1 (47), pp. 31–35.
2. Sandler I. L. Vestnik transporta Povolzh'ya [Bulletin of Volga region transport]. 2012, no. 2 (32), pp. 68–74.
3. Katsyuba O. A., Sandler I. L. Sistemy upravleniya i informatsionnye tekhnologii [Control systems and information tehcnologies]. 2012, no. 1 (47), pp. 31–34.
4. Chen H. F. Stochastic Approximation and Its Applications. Dordrecht: Kluwer, 2002, 357 p.
5. Computer software state registration certificate № 2012615960. Recurrent estimation of parameters being multidimensional in terms of input and output of various orders of a linear dyamic system with noises in input and output signals. Sandler I. L.29.06.2012.
6. Computer software state registration certificate № 2013619621. A recurrent estimation algorithm for parameters of multidimensional linear dynamic systems of various orders with stationary white noises. Sandler I. L. 11.10.2013.
7. Computer software state registration certificate № 203611203. Recurrent estimation of parameters being multidimensional in terms of input and output of various orders of a linear dyamic system with autocorrelated interferences in input and output signals. Sandler I. L. 09.01.2013.