Study On Inverse Kinematics Of Stacking Robot Based On Genetic Algorithm Optimize BP Network

In this paper, under the background of the national development strategy "Made in China in 2025" 's putting forward, along with the growing pressure of the upgrade of Chinese industrial structure, highlighting the problems of the decrease of the demographic dividend, and the demand of stable quality and high value-added processing and manufacturing, stacking robot is been more and more widely used in manufacturing areas, and has become an integral part of the core automation equipment.Currently, stacking robot research mainly in three aspects, including kinematics, dynamics, trajectory planning and control, and kinematics research is the basis of dynamics, trajectory planning and control. Kinematics research mainly including the establishment of motion model and the solution of kinematics equations in positive and inverse ways. The process of inverse solution and trajectory planning is directly related to kinematics analysis, trajectory planning and real-time control etc, therefore, the process of inverse solution is a very important subject in the study of kinematics.This paper take the stacking robot of Xi 'an YinMa Development Industrial co., LTD. as an example, analyze the institutional characteristics, establish the mathematical model based on D-H method so as to seek out the mathematical expression of inverse solution is to lay a theoretical foundation for the subsequent simulation.The kinematical equation of palletizing robot is a very complicated system of nonlinear equations, and the traditional method has a lot of difficulties, via the great nonlinear fitting capability of BP neural network it can solve the solution problem of inverse kinematics equations better. Aiming at the existence of the shortcomings of traditional BP algorithm such as slow speed, low accuracy, easy to fall into local minimum value, using the genetic algorithm to optimize the BP neural network weights and thresh old and thus for the inverse kinematics solution, the result of simulation shows that precision of neural network is obviously improved after optimizing genetic algorithm. Aiming at the problem of premature convergence existing in ordinary single population genetic algorithm, the simulation results show that the precision is further improved by introducing the migration operator and artificial selection operator and optimizing the BP neural network by multiple population genetic algorithm in inverse kinematics solution.The research results show that the proposed using genetic algorithm to optimize the BP neural network method for the inverse kinematics solution is feasible and the simulation results can meet the requirements of the inverse solution.