K. Qian, Y. Zhang, University of Kentucky, Lexington, KY
In Plasma Pipe Welding (PPW) process, the peak current (Ip) is used to establish the keyhole and the base current (Ib) is used to sustain the keyhole for a pre-specified period (keyhole sustaining time) and reduce the heat input. The keyhole-establishment time (Tp) and the keyhole sustaining time (Tb) are controlled by Ip and Ib. When Ip is too low, the keyhole will take a long time to establish; but Ip goes too high burn-through may occur. Hence, the keyhole establishment and keyhole sustaining time need to be controlled at desired levels. In this work, the model predictive control (MPC) method is used to control the outputs (Tp and Tb) so that they track their reference values by calculating the needed inputs (Ip and Ib) using the output feedback in PPW. This requires a model be identified to describe the process dynamics based on which the MPC can calculate Ip and Ib from the output feedback. To this end, a linear structure model was chosen for the process. The Pseudo-Random Multi-level Signal (PRMS) was input into the process to generate the input-output data and the Least Squares (LS) method was used to identify the parameters of the model from the acquired input-output data. The parameters were also identified online to avoid the model deviation. Using the identified model to simulate the outputs, the comparison between the simulated outputs and the measured outputs shows that the linear model matched the PPW process well. Results of closed-loop control experiments showed that Tp and Tb tracked the reference values accurately. Even when the initial model parameters did not match those of the real process, the online identification was able to modify the parameters and the resulting Tp and Tb converged to the reference values.
Summary: Model Predictive control is used in the arc control of pipe plasma welding. The process and the method are introduced and the experiments show the result of the control.
