Mescheriakov D. Models and methods of information processing in biotechnical homeostatic systems

The dissertation is devoted to the development of models and methods of data processing in information systems containing biological components, input and output information of which in real time of the process participates in the functioning of the system. The urgency of the problem is due to the fact that the principles of data processing and decisionmaking by technical and biological objects are significantly different, and the need for systems in which man is an integral part of it is growing. The deterrent is the need to take into account the requirements for the individuality of the initial features, the fuzziness and ambiguity of the reaction of a biological object to the input test effect. The research was carried out on the basis of an infrared biotechnical system, as an indivisible component a person is involved, whose physiological reactions to the input effect are the primary data for processing. The expediency of primary processing of feature information by correlation methods, the parameters of which are consistent with the characteristics of the biological object, followed by filtering with an exponential Brown filter, is shown. The use of non-digital statistics methods for data processing using the Kemeny median to determine the mean in the window, which reduces the impact of uncertainty in the transformation of information. On the basis of non-digital statistical methods, nonlinear filters were implemented, which provided higher resistance to anomalous emissions compared to linear filters. Further processing of the data by a fuzzy controller allowed to reduce the impact of the scatter of the reaction of biological objects on the same impact, which together with the correlation and non-digital data conversion contributes to the creation of systems with biological feedback.