Forward kinematics of parallel robot based on neural network Newton-Raphson iterative algorithm

摘要

A novel redundant actuation 2RPU-2SPR parallel robot is designed. It is a five-axis hybrid machine tool which can be used for complex structure processing in the fields of large ships, new energy and so on. It is composed of parallel mechanism and series mechanism. The combination of the two parts combines the advantages of serial and parallel robots together, enabling the whole machine to have high stiffness, speed, dynamic performance and complex surface treatment capabilities. Next, the forward kinematics was solved, that is, When the parameters of the driving element are known, the parameters of the end-effector are calculated, and the BP neural network optimization strategy is adopted In order to improve the accuracy of BP neural network optimization strategy, the reverse kinematics of the parallel mechanism was derived in detail, and the BP neural network optimization strategy was combined to generate position compensation, so as to achieve high efficiency and high convergence rate. By analyzing the simulation results of parallel robot, it is concluded that this method is suitable for solving the forward kinematics of parallel robot.