Dynamic Error Analysis And Compensation Of Industrial Robot Caused By Flexibility

With the development of industrial robots in the engineering technology such as lightweight, high-speed, heavy-duty and high precision, it is attracting more and more attention that the study of robot dynamic error caused by flexibility. It can’t meet the requirements of modern industrial robots that regarding the joints and links as rigid body. Dynamic error compensation system based on robot dynamics model is difficult to control the robot in real time, staying mostly at the level of theory. This paper reversely builds robot error compensation model based on BP neural network using the dynamic error data obtained by simulation to compensate dynamic error.Firstly, build a rigid coupling model of the robot, do the modeling and simulation of kinematics and dynamics, analysis the effects of different loads to the forces and moments on the robot’s each joint, obtain the theoretical end position data. Secondly, build the flexible coupling dynamics model; discrete the Lagrange dynamic equation of robot overall system using assumed mode method; do the numerical simulation for dynamic equation; get end position data of rigid coupling model of the robot; do the co-simulation by ANSYS and ADAMS software. Sixth-order mode of the waist joint, arm, shoulder joint, the arm is imported for ADAMS software and then rigid coupling simulation model will be built. Analyze force and moment of each joint and do a comparison with the rigid-body model. Dynamic error is extracted from the kinematics simulation under different joint acceleration and load.At last, the dynamic error is processed into input sample data and target sample data. Dynamic error compensation system is built by training the sample data in BP neural network. The errors between resulting output from new sample simulation and target output is very small. It verifies the correctness of compensation and achieves good results.