Evaluation of the injection molding process behavior during start-up and after parameter changes using dynamic time warping correspondences

Abstract

Injection molding is one of the most widely used plastic processes due to its mass production capability and low production costs. Both simple and complex part geometries can be produced in short cycle times which, with other manufacturing processes, would require long production times. To control the process, it is adjusted by a variety of parameters. In addition to the machine configurations, disturbances, the thermal behavior of the machine components and the process control influence the process behavior and the resulting part quality.Due to the large number of machine parameters and the complexity of the process, it is not apparent to the machine operator how long the machine needs after switching on until the process runs stable. Up to now, the quality characteristics of the injection molded parts are determined by extensive tests in order to evaluate the process behavior. Thereby mostly the weight is used representatively as quality characteristic for process and quality control because it can be determined non-destructively, and it is closely related to other quality characteristics (such as dimensional properties).The aim of this research was to analyze the time series of the machine process data and to assess an automated method for evaluating the start-up behavior and the machine behavior after machine parameter changes. For this purpose, dynamic time warping correspondences (DTW) was used to investigate the process behavior based on high-resolution process data (here: injection pressure, flow rate, and screw volume).The method presented in this paper makes a significant contribution to automating the start-up process and reducing the testing effort. In addition, the investigations provide insightful findings for analyzing the process behavior after parameter changes. To improve the outcome, the mold temperature and cavity pressure should also be evaluated with the presented method in future investigations.