The Research On Immune Artificial Algorithm And Its Application In Pose Analysis And Error Compensation Of Robot

One of the obvious characters of modern science and technology's development is that life science and engineering science promoting, permeating and crossing each other. As we known, Biological immune system contains plenty of information processing information processing mechanisms. Through researching, we can design many models and algorithms to solve many complex problems, and it is of great significant to many engineering technology researching.Based on the domestic and foreign research status, further research and improved algorithm of the artificial immune algorithm is given in this paper. And the improved algorithms used to realize the pose analysis and error compensation on robot system.First, considering that the information entropy has the performance to quantitatively evaluate the similarity of distribution characteristic and variance information between the different random variables, we introduces information entropy in immune algorithm to quantitatively evaluate the fitness and density of antibody. This new method can overcome the disadvantage of traditional rules,such as Haiming rules and r-matches rules, which have poor knowledge presentation ability and lack the actual physics significance.Secondly, based on the advantage of artificial immune algorithm which has parallel random search features, we combine the artificial immune algorithm and the BP neural network together to be a immune neural network to solve the problem that BP neural network algorithm easy to fall into local optimum and immature convergence. The neural network weights in the immune neural network can be optimized to improve the global search ability of the network by using artificial immune algorithm as designed in this paper.Finally, the proposed algorithms are used in inverse kinematics solution and position error compensation on manipulator. The information entropy based immune algorithm is used to do inverse kinematics solution of robot system and immune neural network is used to compensate the positioning error of parallel manipulator, all of which effectively resolves the problem of manipulator collision and overcomes the shortcomings of huge calculations and no solution when using traditional mathematical method. The validity of the methods is shown by simulation.