Radial error feedback geometric adaptive control for bar turning in CNC turning centers

In this research a new approach, radial error feedback geometric adaptive control (REFGAC), was proposed to overcome the problems in geometric adaptive control of CNC machines. First, a non-contact laser sensor was used to directly measure the radial error of the workpiece while it is being cut. The sensor is easy to install and calibrate. Accuracy analysis was performed to ensure that the sensor can maintain high accuracy in a rough cutting environment. Second, inspired by the open modular architecture control (OMAC) concept, the X axis controller is replaced by a generic PC. Due to the versatility of the PC, openness of the conventional CNC controller is greatly improved. Not only changing the cutting tool position in real-time becomes possible, but also different control strategies can be easily realized on the PC platform. Third, two modeling techniques, namely, analytical modeling (using classical beam vibration theories) and finite element modeling (FEM), were employed to develop simple models for bar turning process. The analytical modeling is only suitable for uniform bars while the FEM can be used for both uniform and non-uniform bars. The parameters in these models can easily be obtained. Finally, to compensate for the inherent measurement delay, Kalman filters for prediction together with PID control laws were designed based on these two models. Experimental results show that the radial error of the workpieces machined with REFGAC was reduced by 95%.