Research On The Key Issues In Rapid-Precision Motion Control Of IC Bonding Motion Stage Based On ELM

With the continuous improvement of IC chip integration,miniaturization has become a direction of development.The size of the chip and the space of wire bonding& pad pitch are more and more small,improving a higher requirements on the precision and running speed of the packaging equipment.In the circumstances,how to completely overcome the nonlinear friction in high precision control,has become a huge challenge for control approach of high acceleration system.In the process of high-acceleration motion,the nonlinear friction of IC packaging robot presents complex friction phenomena in different motion states.This is the main factor that affects the performance of macro and micro motion stage.The iterative learning algorithm does not need the accurate mathematical model of the system.Only the input and output signals can be used to realize the position control of high precision and high dynamic response by repeated learning,and the repetitive nonlinear friction can be overcome.However,it is a general problem to be solved which is the low efficiency by iterative learning for any new trajectory.The paper firstly introduces the basic principle of iterative learning controller,and the iterative learning control based on sensitivity function identification.By iterative learning for the various programmed training trajectories,we analyse the relative physical quantities with nonlinear friction though the iterative learning data.Secondly,data driven based on the idea of the extreme learning machine(ELM)combined with iterative learning data which has been confirmed can precise tracking for a specific trajectory and contains rich features of potential characteristics of nonlinear friction,a nonlinear friction model based on EI-ELM algorithm has been designed,which shows the advantages of ELM algorithm about high accuracy and precision,and has been made full use of ILC data.Finally,through the combination of Matlab GUI and VB.NET,the algorithm of,system identification,the design of iterative learning law,trajectory tracking,data storage,and EI-ELM nonlinear friction model have been connected to the PMAC motion control card,to compiling an integrated software for simplify user operation and compensate nonlinear friction by control on panel preparation,also promoting the application of iterative learning level in industry.The research of this paper is of great significance to improve the dynamic and static performance of IC packaging robot under the limit condition,and to improve the level of industrial application of iterative learning control.