Research On Accuracy Compensation Method Of Industrial Robot Based On ELM

In recent years,industrial robots have been widely used in various fields of high-precision work.Due to the difference between the parameters of the actual kinematic model and the nominal parameters of the robot,as well as other factors in the process of motion,the positioning error of the robot will increase and can not meet the requirements of high precision work.Therefore,in order to solve this problem,ABB IRB120 robot and SCARA robot are taken as the research objects,and extreme learning machine(ELM)is used to compensate the positioning error of the robot,and on this basis,the ELM compensation model is optimized.The main contents of this paper are as follows:(1)The error model is established based on the kinematic theory of the robot.The positive kinematics model of the IRB120 robot and the SCARA robot is established by the D-H method,and the inverse solutions of the IRB120 robot and the SCARA robot are obtained by using the analytic method.The error model of the IRB120 robot and the SCARA robot is established.lays a theoretical foundation for the precision compensation experiment of the robot.(2)This paper presents a robot precision compensation model based on ELM algorithm,which takes robot joint angle as input,error of actual position and theoretical position as output.The simulation experiments on accuracy compensation of IRB120 robot and SCARA robot are carried out respectively,and compared with the traditional BP neural network compensation model,the results show that the compensation efficiency of ELM model is much higher than that of BP neural network,and the compensation accuracy is higher.(3)In order to improve the stability and compensation accuracy of ELM precision compensation model,genetic algorithm(GA)and particle swarm optimization algorithm(PSO)are used to optimize the input weights and threshold of ELM model,respectively.The GA-ELM model and PSO-ELM model are established,and the simulation experiments of precision compensation for IRB120 robot and SCARA robot are carried out respectively.the results show that the compensation effect of PSO-ELM model is better than that of GA-ELM model.In order to further improve the performance of PSO-ELM compensation model,the improved PSO algorithm is used to optimize ELM,.The results show that the improved PSO-ELM model has better compensation effect.(4)The accuracy of the zero position of the robot directly determines the absolute accuracy of the robot.In this paper,a zero calibration method of SCARA robot based on calibration board is adopted,and the same calibration hole is inserted into the same calibration hole in different positions by the SCARA robot.The zero position information is calculated,and the zero calibration experiment is carried out.(5)In order to verify the compensation effect of the robot precision compensation model established by(2),(3)in practical application,an experimental platform for robot precision compensation is built based on SCARA robot and trilinear coordinates measuring instrument.The unity of robot coordinate system and measurement system coordinate system is completed by using trilinear coordinate measuring instrument,and the method of measuring robot end position is worked out.Finally,through the precision compensation experiment,it is concluded that the effect of the improved PSO-ELM precision compensation method is better than that of other methods,and the absolute positioning error of the SCARA robot compensated by the improved PSO-ELM algorithm is reduced by 80.14%.