Plastic Injection Molding Adaptive Optimization And Learning Control Technology

Product precision,energy saving and high efficiency are the main directions for the development of plastic injection molding.The optimization and control technology of the injection molding process is the key to achieve the above objectives.Artificial intelligence technology,especially adaptive optimization and learning control,has developed rapidly in recent years.It provides a new solution for injection molding optimization and control.This paper studies adaptive optimization and learning control methods.Making full use of batch repetitive characteristics of plastic injection molding process to improve the forming precision and efficiency,and reduce the energy consumption of injection molding process.The main innovations are as follows:Aiming at the problems of long heating time and large temperature overshoot in the process of injection molding heating,a precise control method of barrel temperature with rapid heating process is proposed.Firstly,a prediction control method based on historical data is proposed for the primary heating process from normal temperature.Secondly,On the basis of predictive control method,a new control strategy combining feedforward compensation control and predictive control is proposed for the secondary heating process from high temperature state.The experimental results show that,in the primary heating process,compared with the conventional control method,the proposed method can reduce the temperature overshoot by about 50%,and can shorten the heating process time by more than 20%.In the secondary heating process,the time of the entire heating process can be shortened by more than 60%,and the maximum temperature overshoot is reduced by more than 85% compared with the conventional control method.Aiming at the problems that the temperature fluctuates greatly during the dynamic plasticizing process and the consistency of the final formed products is poor,a precise control method of the melt temperature in the dynamic plasticizing process is proposed.A new dynamic adaptive temperature compensation controller is designed.Firstly,a dynamic compensation time self-learning method based on data is proposed to realize the automatic calculation of compensation time.A self-learning method based on deep reinforcement learning is proposed.The method achieves precise control of the amount of compensation.Furthermore,A self-learning method based on deep reinforcement learning for dynamic compensation is proposed,which realizes the precise control of compensation amount.The experimental results show that the proposed control method can effectively reduce the temperature fluctuation,and the maximum temperature fluctuation is only ±0.5?,which is much better than the PID control method(±4.5?)and the GPC control method(±2.5?).At the same time,the consistency of injection molding products has also been significantly improved.The statistical analysis of the experimental results shows that the product repeatability is less than 0.3% when using the proposed method,which is significantly better than the PID control method(about 1%)and the GPC control method(about 0.7%).Aiming at the traditional quality control methods based on process parameter optimization,which focused too much on quality and ignored the issue of forming energy consumption.An adaptive optimization method for plasticizing process with optimal forming energy consumption is proposed.The proposed method first analyzes the relationship between process parameters and energy consumption during injection molding process.And an evaluation criterion for characterizing melt plasticization using injection pressure is proposed.Based on the evaluation criterion,a parameter optimization method using Sarsa learning is further proposed to achieve an effective reduction of energy consumption in the injection molding process.The experimental results show that,the proposed method can reduce the energy consumption per unit material by about 10%compared with the traditional artificially obtained process parameters.At the same time,the proposed method can well guarantee the plasticization quality of the melt and the quality of the final formed product.Aiming at the problems of low efficiency and poor positioning accuracy in mold opening motion control.Firstly,a trajectory self-planning method using deep Q-learning is proposed to obtain the optimal motion trajectory.After obtaining the optimal mold opening motion trajectory,a trajectory tracking controller combining iterative learning control and feedback control is designed to achieve accurate tracking control of the motion trajectory.The experimental results show that on the basis of ensuring the stability of the mechanism,the proposed method can reduce the time of the mold opening process by more than 35%.Meanwhile,the proposed method can still accurately tracking the motion trajectory when there are repeated and non-repeated random disturbances existing in the system.And the average absolute error of the mold opening position is less than 0.1 mm.The integrated application of the proposed method is realized,and the effectiveness of the proposed plastic injection molding adaptive optimization and learning control method is further verified.Corresponding integration schemes are proposed for the injection machine control systems of EST and HNC.And demonstration applications have been implemented on injection molding machines of brands such as Borche and Chen Hsong.Meanwhile,the effectiveness of the proposed methods are verified by multiple application cases on different injection molding machines.