Identifying and Tuning Mechatronic Systems with State Controllers, Using an Artificial Neural Network

Abstract

This study examines the problem of automatically identifying and tuning control systems for parametrically undefined mechatronic objects with a state regulator. An automatic tuning method is proposed, based on identifying the parameters of a controlled object vector-matrix model, using an artificial neural network, with subsequent synthesis of the controller by the modal control method. An algorithm has been developed to optimize the placement of sensors for controlled object state coordinates, using the mathematical apparatus of observability gramians, which ensures specified identification accuracy in conditions subject to noise in the measurement channels. The proposed intelligent method ensures identification of controlled object parameters and calculation of the state controller according to the results of a single experiment, thus reducing the duration of setting in real time