Abstract
Resin Transfer Molding (RTM) is a manufacturing technology of polymer composites with high mechanical strength and light weight. Its manufacturing procedure includes injecting liquid thermoset resin and hardener into a closed mold to impregnate fiber reinforced metal placed in advance, together with a curing process. In RTM, the race-tracking phenomenon often occur during mold filling, which may cause the formation of dry spot and air entrapment. Because of the irregularity in fiber arrangement, the resin flows toward the path with least resistance during the manufacturing process, leading to insufficiently impregnated defects in low permeability zones. One way to improve quality of product and reduce the possibility of generating defects during mold filling is flow control. Previous research indicates that a constant flow velocity is beneficial to good product quality. However, the existing flow control strategies for RTM are often based on an impractical assumption that the permeability of the fiber reinforcement is uniform within the mold and known in advance. The traditional proportional-integral-differential (PID) controller is unsuitable either, because of the time-varying characteristics of the process. To solve these problems, the recursive least squares method with an adaptive directional forgetting factor is adopted in this study to identify the time-varying process model in real time. Then, a self-tuning PI controller is built based on the identified model for constant flow velocity control, whose parameters are adjusted online according to the Ziegler-Nichols continuous cycling method. In addition, a model predictive control (MPC) strategy is also proposed which controls both the shape and velocity of the resin flow front simultaneously. In order to apply the control structure to on line flow control of RTM, this study adopts mold with single injection gate and three injection gates, let resin flow through low permeability region in model filling, and compares the control performance of tradition PI controller, self-tuning control PI controller and optimizing controller in RTM.