Controle de força e rotação de uma unidade de reparo por atrito usando controlador PID e inteligência artificial

Abstract

This work presents the development of a new method for force and rotational speed control of a machine that performs Friction Hydro Pillar Processing (FHPP) tests. The main challenge in this case is the large torque variation that occurs during the FHPP process that may result in instability and a system stalling. In order to develop the new control system it was necessary to restructure the electrical system and to reprogram the machine. Additionally, PID (Proportional Integral and Derivative) and NN (Artificial Neural Network) controllers were used to control the pin axial force and its rotational speed, respectively. To define the rotational speed NN controller, various NN structures were tested, whereby the best performance was obtained using a NN composed of one neuron with a linear activation function. A GA (Genetic Algorithm) made the adjustment of the controllers using mathematical models obtained from the system identification and FHPP machine itself. The GA was developed to perform the tuning taking into account the integral time absolute error, overshoot and settling time, which were evaluated from a methodology that uses the fuzzy logic. An anti-stalling system was developed to reduce the probability of locking up the rotation during the test and a methodology for rapid depressurization of the hydraulic system was introduced to reduce the fluid hammer effect that could damage the machine. Finally, FHPP tests were carried out to compare the performance of the new control system with the previous one and to define the new operating range of the machine. The developed system showed a considerably better performance in regard to the force and rotational speed control. The new operational range of the FHPP machine was increased about ten times regarding the previous one.