Smoothing, Optimization, and Learning of Control Actions in the Control Systems of Manipulator Robots with Flexible Elements

摘要

Using spectrum analysis, recommendations for selecting standard acceleration and deceleration laws for manipulator robots with flexible elements are presented. For high requirements on speed of response, it is shown that simple smoothing of the laws of acceleration and deceleration may lead to an increase in the amplitude of the oscillations. A problem of parametric optimization of control actions, generated using the derivatives of the standard acceleration and deceleration laws under the assumption that the transfer function possesses dominant complex-conjugate poles, is solved. An iterative procedure that enables the control actions on a practical object to be optimized in a preliminary accommodation mode is constructed.