TY - CONF

T1 - Application of Data Mining for Digital Twin and AI-based Process Optimisation in Hot Forging

AU - Peddinghaus, Julius

AU - Brunotte, Kai

AU - Wester, Hendrik

AU - Behrens, Bernd-Arno

AU - Rothgänger, Marcel

AU - Hinz, Lennart

PY - 2025/10/7

Y1 - 2025/10/7

N2 - The technology of forging has continuously been reinvented throughout its long history. Today’s complex forging process chains are characterized by their high volatility and complexity in interacting mechanisms. To manage these challenges, automation has consistently been implemented further throughout the last decades to increase output, reproducibility and quality. To this day, automation is however limited to forging processes with high quantities due to high costs of setting up an automated forging process. The large variety in geometry and process conditions limit the transferability of automation concepts between different forging processes. Forged products without high quantities and long-term commitment are therefore still carried out manually by highly trained and experienced experts. The status quo faces two major upcoming challenges: As product life cycles consistently decrease and the labour market is increasingly strained through demographic shifts, the need for adaptable, flexible forging process automation can no longer be ignored. This raises the research question of the presented work, how forging processes can be optimized through a digital twin approach despite limiting factors such as cost-intensive tooling.In this presentation the challenge of flexible forging processes is addressed through two approaches: data-based process optimisation through artificial intelligence and adaptable, self-controlling process components. The former can provide the basis for the latter. Automated forging equipment inherently requires substantial data to maintain and control a process. So far this has been used to control process sequences but remains largely untapped for further data analytic and process optimisation. Furthermore, recent advances in sensor technologies offer the potential to enhance data acquisition even in robust forging environments. The results presented here show that data acquisition provides the key basis to enhance process parametrisation and design.The aggregation of all data associated with each forged component throughout a process chain to a digital twin offers the additional potential to individually adapt subsequent processes for increased quality. Current processes are parametrised collectively for an entire production run, disregarding dynamic effects. Adaptation is limited to individual experience-based interventions by a process operator. The results presented in this contribution show, that through the combination of data acquisition with adaptable process hardware, subsequent steps can be parametrised individually for each part based on its process history. Through the combination with AI-driven process optimisation, processes can learn and improve with each forging cycle reducing rejects and digitising process knowledge so far confined to each process operator’s experience.

AB - The technology of forging has continuously been reinvented throughout its long history. Today’s complex forging process chains are characterized by their high volatility and complexity in interacting mechanisms. To manage these challenges, automation has consistently been implemented further throughout the last decades to increase output, reproducibility and quality. To this day, automation is however limited to forging processes with high quantities due to high costs of setting up an automated forging process. The large variety in geometry and process conditions limit the transferability of automation concepts between different forging processes. Forged products without high quantities and long-term commitment are therefore still carried out manually by highly trained and experienced experts. The status quo faces two major upcoming challenges: As product life cycles consistently decrease and the labour market is increasingly strained through demographic shifts, the need for adaptable, flexible forging process automation can no longer be ignored. This raises the research question of the presented work, how forging processes can be optimized through a digital twin approach despite limiting factors such as cost-intensive tooling.In this presentation the challenge of flexible forging processes is addressed through two approaches: data-based process optimisation through artificial intelligence and adaptable, self-controlling process components. The former can provide the basis for the latter. Automated forging equipment inherently requires substantial data to maintain and control a process. So far this has been used to control process sequences but remains largely untapped for further data analytic and process optimisation. Furthermore, recent advances in sensor technologies offer the potential to enhance data acquisition even in robust forging environments. The results presented here show that data acquisition provides the key basis to enhance process parametrisation and design.The aggregation of all data associated with each forged component throughout a process chain to a digital twin offers the additional potential to individually adapt subsequent processes for increased quality. Current processes are parametrised collectively for an entire production run, disregarding dynamic effects. Adaptation is limited to individual experience-based interventions by a process operator. The results presented in this contribution show, that through the combination of data acquisition with adaptable process hardware, subsequent steps can be parametrised individually for each part based on its process history. Through the combination with AI-driven process optimisation, processes can learn and improve with each forging cycle reducing rejects and digitising process knowledge so far confined to each process operator’s experience.

UR - https://ifc2025.com/programme

M3 - Abstract

ER -