Fast And Effective Two-dimensional Control For Injection Molding Process

Injection molding is a widely applicable process for manufacturing plastic products. Packing pressure and injection velocity are two key variables in the process to determine the final product quality. As a typical batch process characterized in two-dimensional dynamics at time and batch directions, process control of injection molding requires 2D control algorithms. However, due to the fastness of injection molding process and limits on software and hardware of the controllers, traditional 2D control algorithms are hard to apply because of their complexity. Therefore, for such a fast batch process, how to develop a fast and effective 2D control algorithm that can guarantee good control performance and meanwhile reduce the cost of software and hardware of controller, has been a hot topic for process control recently. In response to this demand, this work studies the fast and effective 2D control algorithm for injection molding process.The original contribution of this thesis is summarized as follows:1. Under the 2D framework, a 2D predictive functional control algorithm is developed by combining the predictive functional control (PFC) and iterative learning control (ILC). The fastness and effectiveness are analyzed, and the tuning guidance is provided. The performance in convergence, disturbance rejection and handling constraints are tested through simulation of injection molding process control. The comparison of control performance between IMC-PID and proposed 2D-PFC verifies that the 2D-PFC outperforms the IMC-PID.2. Considering the cost limit of hardware and software of an injection molding controller, two more simplified control laws are proposed to reduce the online and offline computational load by applying the single base function and two base functions. The packing pressure and injection velocity of an industrial-size injection molding machine are controlled with the simplified 2D-PFC. Experimental results on an industrial injection molding machine validate the effectiveness and practical utility of the proposed control algorithm.3. Based on 2D-PFC with the single base function, a simplified first order plus dead time (FOPDT) model is built. The error between model and process is analyzed and predicted. The predictive functional control based on 2D error prediction algorithm is proposed. The proposed algorithm is more concise and easier to tune parameters. The convergence, disturbance rejection and high-order mismatch of the algorithm is also tested through simulation. Comparison experiments show that the 2D-PFC outperforms IMC-PID with model mismatch.