Methods of Constructing the Neural Network Models of Regulators for Controlling a Dynamic Object with Smooth Monotonous Behavior

Abstract

The paper describes the methods of forming the training samples and constructing the neural network models of regulators by increasing the number of layers and neurons in the case of one-channel and two-channel control of a dynamic object with smooth monotonous behavior. The paper presents results of constructing the neural network model of a regulator for controlling the roll motion of an unmanned aerial vehicle with an autonomous roll channel. We consider an example of constructing the neural network models of regulators for two-channel control of lateral motion for unmanned aerial vehicles with a simplified model.