Injection molding is the most widespread process to produce plastic parts for practically all application areas. In this situation, high throughput and good quality parts are important for a manufacturer to survive on the market. For this purpose, machines are usually set up by skilled personnel to produce good parts at the beginning of a batch.

To date, this happens once at the start of a batch and the machine then runs with constant parameters. However, if the ambient conditions change (e.g. the air temperature due to open gates), currently no reaction is possible and the risk of faulty parts increases. In addition, the settings are chosen based on experience. Whether these correspond to the optimum operating point in terms of resource consumption with good product quality is unclear.

In the ProQuaOpt project, we aim to apply AI methods for automated productivity and quality optimisation of the injection molding process. To achieve this, we plan to develop a product-process quality control loop that will consist of a subsystem to interact with the injection molding machine, a measuring element for quality inspection and an AI-based controller.

In addition to responding to changing environmental conditions, the control loop also enables continuous optimisation of the process in terms of resource efficiency. When necessary, the controller provides small adjustments to keep the process at the optimum operating point. Internal feedback is given to the controller via the quality inspection module. The adjustments will mimic steps that an expert would also perform.

In the ProQuaOpt project, IMMS will combine its expertise in signal acquisition and processing for AI applications as well as the development of AI-based predictive models. We will develop a process-optimising control system for both inline and autonomous operation.

Depending on the input parameters, the control system will optimally adjust the settings of the injection molding machine without affecting quality of the produced parts. The system can also respond to external disturbances. This makes the injection molding process more efficient and results in less defective parts.