Optimal Design of the Control System for an Industrial Robot Using DOE Technique and Regression Model
This paper approaches the optimization of the control system for an industrial robot with 6 axes (degrees of freedom), using design of experiments (DOE) and multiple linear regression models. The design objective refers to the desired trajectory of the end-effector, the aim being to minimize the difference between the desired (imposed) and current (measured) angles in the revolute joints of the robot. The correlation between the imposed trajectory of the end-effector and the corresponding angular motions in the six revolute joints is obtained through the inverse kinematic analysis. The characteristic parameters of the controllers are used as design variables in the optimization. The optimal design is based on the DOE Screening investigation strategy with the Full Factorial design type. This design was chosen in order to evaluate the effect of the factors and of their interaction on trajectory, and the levels of these factors needed to produce an optimal trajectory. By comparing actual data with data after optimization, it shows that the regression function is correct (in terms of goodness of fit). The dynamic model of the robotic system was developed in mechatronic concept, by integrating the mechanical device (designed in ADAMS/View) and the control system (MATLAB/Simulink) at the virtual prototype level. The optimization study is performed by using ADAMS/Insight.