Optimal Profiling Study On Injection Velocity

Quality control is of great importance for polymer processing. Injection molding, as an important polymer processing technology is a multistage batch process, including filling, packing-holding, cooling and plastication. Among them, the filling stage may contribute most to the final part quality. To ensure the uniformality of the product, a constant molding filling speed of the melt-front is anticipated. This thesis, therefore, focuses on the study of the optimal profiling method for uniform mold filling.A set of experiment design is first proposed to collect data, based on which a recurrent neural network model is developed to predict the average melt-front-length, which directly relates to the melt-front-rate, by using the measurement of screw injection velocity. Good performance in prediction is achieved by using on this model. Later, three different approaches are proposed for the optimal profiling of the injection velocity. In approach 1, the injection velocity curve is divided ten fixed intervals along the screw distance. Optimization is conducted to find the best velocity for each stage to ensure a constant melt-front rate. In approach 2, it is similar to that in approach 1, except that the interval number is not fixed. Based on that, both the interval length and interval number are optimized. In the third approach, an intelligent interval division method is proposed. Optimizations are later conducted to search for the optimal interval length. All the three approaches are tested on a number of molds with different shape. Comparisons are given through experiments.