Automation of legacy equipment, multifactorial AI-powered optimization, building a small 300mm fab – Whatever drives you these days, the Factory Automation User Group is here to support you!

The 2022 Factory Automation User Group again was a lively exchange of experiences and ideas to tackle current challenges in semiconductor manufacturing.

SYSTEMA solution experts and factory automation end-users from Nexperia, Linxens, Vishay, and Fabmatics provided in-depth insights into current projects and new approaches in factory automation.

  • Arrival & Registration

  • Welcome & Opening

  • Keynote

    Frank Bösenberg | Silicon Saxony e.V.

  • Maximizing due date compliance, throughput and other partially conflicting objectives using multifactorial AI-powered optimization

    Semiconductor production is a highly complex interplay of human, machine, material and method. With state-of-the art meta-heuristics as baseline for WIP scheduling, we at Nexperia Hamburg are exploring also more advanced methods to optimize production. By combining simulation, combinatorial optimization and reinforcement learning, we have created a powerful testbed to prepare better dispatching and scheduling methods. We discuss how to balance – sometimes conflicting – optimization objectives by establishing a robust and transparent model verification process to support line engineers. A holistic twin of a selected bottleneck production area provides not only data to drive the planning the solution, but also enables transparency and predictive analytics to allow revealing unused production potential.

    This presentation is part of an ongoing research collaborative initiative AISSI “Autonomous Integrated Scheduling for Semiconductor Industry” together with Bosch, SYSTEMA, and other industrial as well as academic partners, exploring how to push the boundaries of production scheduling using recent AI concepts.

    Tobias Sprogies | Nexperia, Hamburg
    Martin Krüger | SYSTEMA

  • Coffee Break

  • Improving transparency of Pick&Place production with RPA

    Legacy equipment often lacks proper automation because it doesn’t provide any digital interfaces other than the operator GUI. Even in this case a high level of automation is possible – using SYSTEMA RPA (Robotic Process Automation). By observing the contents of the tool PC GUI and interacting with keyboard and mouse controls, various tasks can be automated, such as monitoring production parameters, supporting the operator during machine setup or automatic reporting of critical alarms. In this talk, we’ll introduce some concepts and common use cases of SYSTEMA RPA and present the solution installed at LINXENS.

    Stephan Radke | LINXENS, Dresden
    Viktor Bilousov | SYSTEMA

  • Challenges while building the smallest 300mm fab in the world

    Vishay’s strategic decision to build its own 300mm production line in Germany is driven by market growth and cost pressure. In order to get the full benefit of such an new fab and to translate the project into competitiveness Vishay needs to think about new concepts in all business processes. One important aspect is the level of automation that you need to find a compromise between operational cost, investments and management of product complexity. The talk will give you some insights into the thought processes, challenges we face and assumptions we made.

    Tilo Bormann | Vishay Siliconix, USA

  • Discussion & Closing

  • Evening Event

  • Arrival & Registration

  • Welcome & Opening

  • Automatic classification of measurement data

    Vishay Siliconix uses spreading resistance profiling (SRP) to control the dopant profile of epitaxy layers. In cooperation with SYSTEMA an automatic classification of the dopant profile has been developed which is used for detection of misprocessing. To train the used model, machine learning techniques have been used.

    Andreas Gondorf | Vishay Siliconix, Itzehoe

  • Using imaging, data augmentation and AI for part classification in high-throughput RFID production

    During production of RFID inlays, chips are assembled to antennas to shape the final product. This is realized with adhesion process. Occasionally, mis-assemblies happen caused by relative material rotation, displacement, or bonding process anomalies. Because thousands of units are produced per hour on a single machine, a manual inspection is not feasible. Together with SYSTEMA, a novel method to automate a part classification using cloud-based AI techniques is developed. To overcome limited bad part examples and to allow easy inclusion of new product types, a fully automated model building process has been established.

    Kevin Hanka | LINXENS, Dresden
    Rakesh Ganya |

  • Coffee Break

  • Enabling a holistic fab view and improved production planning by simulation, dispatching and real-time intelligence

    How can we make a manufacturing process more efficient? How can we produce more? How can we improve tool utilization? Can we do so without hiring new workers or buying more equipment?

    Unfortunately, such relevant questions are often very hard to answer due to complex production processes. To overcome this, we can make use of simulation and digital twin solutions. By this, we can evaluate different planning strategies like dispatching rules risk-free and analyse the results in real-time.

    In our talk, we present an interplay of three SYSTEMA products: several typical production scenarios are simulated by our Factory State Simulator (FSM). By the automated dispatching system EDDi efficient lot processing is ensured. The improvements achieved by dispatching are analyzed and reported in the real-time digital factory twin named RI Suite.

    Gregor Nordheim | SYSTEMA

  • The journey to extend legacy equipment with RFID capabilities

    Jan Klinger | Fabmatics, Dresden
    Johannes Meihsner | X-FAB, Dresden

  • Discussion

  • Catering