Research Of Predictive Control For Injection Velocity Control On Injection Molding

Injection velocity, which is critical for the quality control of injection molded parts, is one of the most important process variables for the injection molding process. In injection molding, the injection velocity has the characteristics such as nonlinear and time-varying and that make the velocity control a challenging task. Because of the high complexity and nonlinearity, the performance of the traditional PID algorithms is not very well on the control of the injection velocity. At present, on the screw injection velocity tracking control, in most of advanced control algorithms have been researched. A nonlinear model-based predictive control algorithm for the screw injection velocity tracking control, which is easy for implementation, is proposed in this thesis. This algorithm can effectively and simply keep the injection velocity under control and achieve better performance. This algorithm has been analyzed by simulation firstly, and then tested on an energy-saving injection molding machine.The main contents in this thesis include four aspects as following:1. An experimental platform for injection velocity control has been constructed in this project. Besides, to satisfy the real time requirements of the data acquisition and processing, a real-time data acquisition system has been developed.2. In this project, the nonlinear characteristic of the injection velocity has been analyzed. Through a step response test, it is found that the changing rate of the system pressure during the injection phase bears close relationship to the nonlinearity of the injection velocity.3. The piecewise points of the model were determined by the changing rate of the system pressure during the injection phase firstly. Then, the sample points of the screw injection displacement were determined by the former piecewise model. Thereafter, a signal for the model identification was designed and applied for an open-loop identification of the injection phase. The injection phase was subdivided by the corresponding moment of the screw injection displacement sample points. Furthermore, the linear models for every segments of the injection phase were obtained. Finally, the LPV non-linear model with the screw displacement as the tuning variable has been established.4. The LPV non-linear model was implemented on the control of the injection velocity. The linear models at every screw displacement sample points were calculated in real time based on the linearization method. Then, these models are used as the predictive model at the corresponding screw displacement sample points. The injection velocity was closed-loop controlled by the MPC controller which was formed via the predictive model. At last, the controller has been realized and implemented online for the energy-saving injection molding machine.