Método de identificação com inteligência computacional e controle de um manipulador robótico

Abstract

This work presents the identification with least squares, recursive least squares, and recursive least squares with particle swarm optimization RLSPSO of a joint model of a robotic manipulator driven by three phase induction motor. Also presented is the model of the first joint, in state space, found from the dynamics of the manipulator, using experimental data of current and velocity. The identification algorithms are evaluated by the metrics multiple correlation coefficient (R2) and by the root mean squared error - RMSE. The controllers LQI and LQG optimized with the metaheuristics GA and PSO are used, where they find the Q and R matrices of the mentioned controllers. In the identification results, the curves of the training and test stages of each algorithm and the errors of each stage are presented, in addition to the discrete transfer functions found by the algorithms, and for the best algorithms, in this case, the RLSPSO, the continuous transfer function and the model in space are presented. of states. The main contributions of this work are the identification with a new algorithm, RLSPSO, and the improvement of the controllers LQI and LQG optimized with the metaheuristics GA and PSO applied to the model identified by the RLSPSO and dynamic model of the manipulator’s joint. Regarding the controllers, the LQI+PSO presented better results than the others. With this, it can be stated that the results of the identification and the proposed controllers were satisfactory for the application in the manipulator’s joint