Parameter recommendation for injection molding based on similarity analysis of injection molded parts

Injection molding is a widespread manufacturing process for producing complex plastic shapes. The process is controlled by a large number of parameters, which show interactions with each other. Thereby the determination of an optimal operating point during sampling is a resource-consuming and demanding task. The data, which is generated during sampling, is usually not used for subsequent sampling processes. In the context of Industry 4.0, manufacturing companies attempt to optimize production efficiency to save time and resources and remain competitive. Machine learning methods offer great potential to analyze correlations in already generated data in order to predict optimal parameters of the operating point for newly developed parts, which can only be determined with high experimental effort. The aim of the investigations is to provide a method for recommending setting parameters of the injection molding machine for new injection molds. This method is intended to make a significant contribution to reducing the elaborate tests during sampling. For this purpose, a concept was developed that, in the first step, calculates the geometric similarity of the new mold to a known database with part information from molds that have already been sampled (here: 743 parts). In the second step, the new injection mold is assigned to a cluster containing geometrically similar parts. Since the information on the robust operating point is known for each part of the database, this information is used to calculate the operating point for the new part. The applied case-based reasoning method provides the machine operator with a recommended operating point for a new injection mold without the need to carry out elaborate tests beforehand. Since the operating point of an injection molding machine consists of several parameters, the investigations were initially focused on the flow rate, which is one of the key parameters for controlling the injection phase.