Research On The Control Method Of Injection Velocity In The Process Of Injection Molding

As an essential process in plastic production, injection molding is becoming increasely vital due to the energy shortage and the excellent characteristics of plastic products. The injection molding process is consisted of injection, pressure keeping, cooling, plastication etc. And the most important one to keep high quality is the injection process in which the injection velocity becomes a key control variable, so how to control the injection velocity effectively appears to be a crucial problem. Taking injection molding as the research background, this thesis conducts a study on the control method of injection velocity.Initially, related literature on injection molding machine and injection velocity is introduced in brief. Then, the non-linear and time-varying dynamic characteristics of injection velocity in the process of injection molding are analyzed in the dissertation.In the next place, based on some basic knowledge about Neural Network (NN) and inverse system, this paper elaborates non-analytical form of inverse system-Neural Network Inverse System (NNIS). At the same time, some related knowledge such as reasons of bringing up, structures, training and control methods about Neural Network Inverse System (NNIS) has been studied in depth.Finally, on the basis of preceding theory foundation, neural network inverse control method is applied to the control of injection velocity. After explaining the injection velocity mathematic model, this paper analyzes its reversibility and confirms that the injection velocity system is reversible. BP Neural Network and RBF Neural Network are utilized to construct injection velocity inverse system respectively, which makes it available for open-loop control of injection velocity. Based on the open-loop controller, the Neural Network inverse composite controller that consists of additional controller and Neural Network Inverse System is designed, the closed-loop control of injection velocity has been realized and validity of the method is verified by simulation experiment. In addition, this paper conducts a contrast about control results of Neural Network inverse composite control system based on BP Neural Network and RBF Neural Network separately and the simulation results show that RBF Neural Network is more efficient. In the end of the dissertation, the inverse control system based on RBF neural network is tested to verify that the system has good robustness.