Features of different types of automatic extremum seeking systems based on localization method

Abstract

The synthesis of automatic extremum seeking systems for a one-channel object whose model can be represented as a series connection of a nonlinear dynamic component and a static quality function with a distinct minimum or maximum is considered. The possibility of extremum seeking by the localization method involving the use of the highest derivative of the output variable of the dynamic part of the object in the feedback is discussed. Depending on the version of implementation of such feedbacks, two main types of extremal systems are distinguished: one-loop systems, in which data on the static characteristic gradient is sent to the controller, and two-loop systems. In the latter systems, it is suggested to use controllers based on the localization method for preliminary stabilization of processes in the internal loop containing the dynamic part. In the external loop, an integral control law is used and motion to the extremum with specified dynamic quality is implemented taking into account the data on the gradient. Modeling shows the features of the extremal systems under consideration.