Development of Optimization Algorithms in Estimation Problems of Non-Stationary Processes

The main goal of this paper is the development of optimization algorithms in problems of estimating nonstationary processes. To achieve this, it is necessary to solve a number of problems: 1. Determine the dynamic characteristics of an object in the form of equations 2. Determine the allowable trajectories of the system and control 3. Build a mathematical model of the object. 4. Set management objectives; 5. Set the quality function or loss functionality. 6. Construct and simulate the control of a non-stationary object. The predator-prey biological system will be used As a basic model, taking into account the characteristics of effectors and therapeutic controls. Depending on the immune system and the type of disease, different models are proposed that are consistent with the experimental results. The model used in the work was built on the basis of several previous ones, taking into account the clinically observed temporal fluctuations and manifestations of the qualitative behavior of the systems. The task of constructing an optimal control of an object with incomplete information with respect to the quality functional, the set of goals, admissible controls and states is the search for a control belonging to an admissible set minimizing the given loss function, taking into account possible unknown interference, not for one particular system, but for their family, Parameters of elements and characteristics of which are taken from known sets. The work was written with using the method of "extended linearization" and SDC parameterization. The behavior of the system was presented during the passage through the various steps of the control algorithm, various options are shown, depending on these steps. A method for introducing an observer for evaluating deviations is also described. The computer simulation of a single task and its results are shown, which once again speak about the effectiveness of the methods used.