Nonlinear control of an electrohydraulic injection molding machine via iterative adaptive learning

This paper examines the control of an industrial injection molding machine (IMM) for both mold filling and mold packing. The machine is a Boy 50 Ton system operated by hydraulics with electronically controlled valves. A nonlinear model of the system is determined for the filling and packing stages, then verified against experimental data. The model includes several key aspects of the actual machine dynamics including stick-slip friction, time delays, nonlinear valve flow characteristics, and deadzones. A feedforward control input is determined via an iterative learning control scheme. This input is combined with either a feedback strategy or an open-loop strategy for the filling and packing control phases, respectively. The time delay, which exists in both stages, is accounted for by time shifting the learning feedforward signal relative to the feedback control signal. Simulation and experimental investigations indicate the benefits of the proposed strategy for control of electrohydraulic IMM's. Subsequent to demonstrating the benefits for fill-stage and pack-stage learning control, a complete cycle, including fill-to-pack transfer, is demonstrated.