Data-driven Based Controller Parameters Tuning For Wire Bonders

With the development of modern control theory, more and more types of controllers are applied to the practical industrial control system. However, near 90% of them are PID controllers. Since there is no efficient way to tune these controllers, most of them are tuned by hand, which greatly reduce the efficiency. Moreover, we can't get the optimal solution by this way and the parameters are not robust.This thesis proposes to use the data-driven methods to tune the parameters for the controller of LED golden wire bonder. First, the non-iterative data-driven method is used to do the coarse tuning for the system and then the iterative data-driven method is applied to fine tune the wire bonding system. Finally, the optimal parameters for the system are achieved. The non-iterative data-driven method used in this thesis for coarse tuning is the non-iterative correlation based tuning method. It is compared to the virtual reference feedback tuning method to prove its good performance. The iterative data-driven method used in this thesis for fine tuning is the iterative correlation based tuning method which is comparing to the criteria based data driven method.Simulation is made in MATLAB and shows the applicability of these methods. Then, these methods are applied to the LED golden wire bonder system to get the parameters of the controller. The experiments for the VRFT and NICBT methods are implemented on the XY table of the golden wire bonder. I can get applicable parameters for the controller with fast speed. The result of NICBT method is used as the initial parameters for the controller for fine tuning and the ICBT method is used to get the optimal solution near the solution of NICBT. The coarse tuning plus fine tuning mode can get a good result when applying to the LED golden wire bonding system.