Optimal Tracking Control of Injection Speed in Injection Molding Machine with Gaussian Process Learning and Model Predictive Control

Injection molding, as a common plastic processing technology, has significant importance in modern manufacturing industry. However, due to the nonlinear dynamic characteristics of injection molding machines (IMMs) and the influence of various factors such as raw materials, equipment, and environment during the injection molding process (IMP), uncertainty residuals are generated, which makes it difficult to achieve optimal tracking control of injection speed, thereby affecting product quality and production efficiency. To address this issue, we propose a learning-based model predictive control (LMPC) scheme. The scheme uses Gaussian process regression (GPR) to predict and model uncertainty residuals in the IMP. The residual model of IMM is learned through historical injection molding data, and the model predictive control (MPC) is used to achieve optimal tracking control of injection speed based on the predicted model, thereby automatically adjusting the injection speed according to requirements. In addition, we use differential dynamic programming with control constraints to solve the optimization problem. Finally, we conduct comprehensive numerical simulation experiments to prove the effectiveness and efficiency of the proposed LMPC scheme in injection speed control of IMM.