A comparison of model predictive control and PID temperature control in friction stir welding

Abstract

Temperature control of friction stir welding (FSW) via model predictive control (MPC) is investigated in Al 7075-T7. Two MPC controllers are compared against two well-tuned PID controllers to obtain a direct comparison of MPC and current FSW controllers. One MPC controller uses a first-order plus dead-time (FOPDT) model derived from a simplified conduction-advection view of the stir zone. The other MPC controller uses the Hybrid Heat Source model that describes heat conduction in the plate and tool.At quasi steady-state conditions, all four controllers can easily hold temperature within 2°C of the setpoint in the absence of large disturbances. Once the weld is past the initial traverse, the FOPDT controller is superior to the Hybrid Heat Source controller with regards to modeled-disturbance rejection and setpoint changes. The FOPDT controller is competitive with well-tuned PID controllers in this region of the weld. During the initial traverse, the Hybrid Heat Source controller and PID controller with regulator gains were able to control temperature within 5°C of the setpoint, compared to a typical deviation of 20–30°C when uncontrolled. During this period, the FOPDT controller and PID controller with servo gains could not maintain satisfactory temperature control.MPC is demonstrated to be a viable control method for FSW. Temperature control before reaching steady state for both MPC and PID is shown to be feasible, but more difficult than for steady state. Recommendations are given for when each controller might be preferred in various circumstances, based upon the results shown herein.