A Dynamic Programming approach for batch cycle time optimization in hot metal forming*

Abstract

In the framework of the Fourth Industrial Revolution, the manufacturing industry has been immersed in a digitalization process leading to constantly increasing available data at the production line thanks to novel sensors, Cyber-Physical Systems, and the Industrial Internet of Things. These newly available process data streams can be leveraged for realtime planning, control, and process optimization towards a more efficient and competitive manufacturing paradigm. The metal forming process is a manufacturing process used in a wide variety of complex industrial pieces and products: from automotive parts such as car body panels to furniture, electronics, etc. Ensuring high quality and robust materials while optimizing production of parts is an important industrial challenge. Nowadays, batch and cycle time optimization rely on experience and expert knowledge. In this work, we propose a dynamic programming approach for real-time decision-making in hot metal forming production to minimize batch and cycle time. This paper introduces a successful proof of concept, and the first step for an autonomous self-learning control system in stamping processes, demonstrating its capabilities for overall batch and cycle time process optimization while maximizing product quality.