NEURAL NETWORK MODELS OF ADAPTIVE POSITION-TRAJECTORY CONTROL SYSTEMS OF MOVING OBJECTS

This article is devoted to neural network models of adaptive position-trajectory control systems of moving objects. Neural network models of adaptive positional-trajectory control systems of moving objects have been developed on the basis of proportional-differential adjusters, which allow for quick calculation and determination of nonlinear characteristics of the system. Also, the kinematic scheme of the multi-link industrial robot is presented for describing the motion of the multi-link industrial robot according to the specified traction and positions, and for representing the input forces and training in the neural network. Mathematical models for calculating the linear and rotational movements of the manipulator handle in the x and y coordinate positions and directions are presented. The presented kinematic scheme and mathematical models serve to develop the structure of adaptive position-trajectory control systems of multi-link industrial robot manipulator based on neural network models. In the control structure, the two-layer model of the neural network allows calculation of tracking errors and moments in the specified coordinate trajectory, individual control of each link based on the signals received from the position sensor of the links, and visualization based on computer technologies.

Перевод пока недоступен