Reinforcement learning supported quality control loop for solid forming processes

In solid forming, the ramp-up phase of new batches poses a challenge due to process instabilities and frequent dependence on expert knowledge. Reinforcement Learning can be used to mitigate process instabilities and lead to a quicker fulfillment of the required quality characteristics. A quality control loop (QCL) based on reinforcement learning (RL) has been developed to determine the optimal process parameters for solid forming processes. The QCL continually provides recommendations for process parameters based on quality characteristics. The validation and benchmarking of the QCL is carried out based on a use case in the solid forming industry.