Fedorova H. Method and tools of information technology for nonparametric dynamic models identification of the oculomotor apparatus.

The thesis is devoted to actual scientific and technical problem of mathematical software development methods and information technology nonparametric identification of dynamic models of oculomotor apparatus (OA) rights. Subject study is due to the contradiction between the increasing requirements for accuracy of models describing the OA, on the one hand, and increasing speed identification procedure on the other side. This contradiction is resolved using information technology for the identification of nonparametric dynamic models in form of integro-power Volterra series using multidimensional weighting function (MWF) and multidimensional transition functions (MTF) that can simultaneously take into account nonlinear and dynamic properties of the OA and reduce the amount of computing. The purpose of the study is to improve the accuracy and speed of constructing mathematical models of oculomotor system as integro-power Volterra series through the development of effective methods and tools of information technology nonparametric identification of dynamic objects. To achieve the purpose of the study raised and solved the following problems: – analyzing the existing methods of identification of continuous nonlinear dynamic objects and justify the choice of research direction to build integrated nonparametric dynamic models based on MWF and MTF to describe human OA; – analyzing the current state of reduction techniques and information models to justify the choice of research direction to reduce the dimensionality of models based on correlation filtering methods; – developing the method of nonparametric identification OA with MTF based on the results of these active experiment "input-output" using test polypulse and multistage signals to ensure acceptable accuracy of the models; – developing the information technology for identification of nonlinear dynamic objects, based on the OA description on the basis of nonparametric dynamic models in the form of MWF and MTF, reduction of constructed models based on correlation filtering methods; – applying the developed method and means of information technology of identification of continuous nonlinear objects for construction of nonparametric dynamic OA models.