Model-Predictive Control of Servo-Pump Driven Injection Molding Machines

A high-dynamic and accurate control of the process variables within their admissible limits is essential to meet a high product quality in injection molding processes. Conventional control concepts of hydraulic injection molding machines (IMM) typically employ servo-valves or variable displacement pumps as actuators. Nowadays, servo-motor driven pumps are frequently used due to their higher energy efficiency. These systems, however, exhibit a slower actuator dynamics and, thus, demand for more advanced control concepts. This paper proposes a novel control concept for this type of IMM consisting of a Lyapunov-based load volume flow estimator and a model-predictive controller based on a Riccati recursion. The control concept systematically accounts for the system constraints and features a high performance for the filling and packing phases without knowledge of the mold geometry or information from the previous injection cycles. A thorough comparison of the proposed control strategy with the industrial state-of-the-art controller is made by injection and melt cushion experiments on an industrial IMM. Finally, also the robustness of the proposed control concept with respect to model uncertainties is shown.