Analysis of heterogeneous and synthetic data driven artificial intelligence for exploring smart electromagnetic forming (AI4EMF)

分析异构和合成数据驱动的人工智能，探索智能电磁成形 (AI4EMF)

• 批准号: 513332912
• 负责人: Professor Dr.-Ing. Steffen Ihlenfeldt
• 金额: --
• 依托单位: Fraunhofer-Institut für Werkzeugmaschinen und Umformtechnik (IWU)
• 依托单位国家: 德国
• 项目类别: Research Grants
• 资助国家: 德国
• 项目状态: 未结题
• 来源: https://gepris.dfg.de/gepris/projekt/513332912?language=en
• 关键词: Analysis; heterogeneous; synthetic; data; driven

Electromagnetic forming (EMF) has high potential for shaping complex parts and implementing lightweight design concepts, but despite of advantages such as high flexibility and extended forming limits at room temperature for many materials EMF has not made its major breakthrough yet. One reason is that due to complex interactions between thermal, electrical and mechanical field variables, it has hardly been possible to design a manufacturing process including the necessary durable tool coils in a simple and efficient way. Furthermore, due to the very high forming speed hardly any reliable conclusions about the process conditions can be made. Current developments in computing power and new technologies such as live process data analysis, cloud computing, connectivity and controlled production, enable overcoming these challenges. The planned project combines developments in the field of artificial intelligence (AI), numerical modelling and process technology to enable the investigation of a smart EMF process. Data-based AI methods shall ensure systematic process design and control. For this purpose, synthetic data from simulation and domain knowledge are consulted together with conventional process data in a technologically barely explored approach. Additionally, a method to design durable tools for EMF based on comprehensive numerical modelling is developed. Finally, all results are combined and tested based on a demonstrator process.

电磁成形（EMF）在复杂零件的成形和轻量化设计概念方面具有很大的潜力，但尽管具有高灵活性和在室温下对许多材料进行扩展成形限制等优点，但EMF尚未取得重大突破。其中一个原因是，由于热、电和机械场变量之间复杂的相互作用，几乎不可能以一种简单有效的方式设计一个制造过程，包括必要的耐用工具线圈。此外，由于成形速度非常快，很难得出任何可靠的工艺条件结论。当前计算能力和新技术的发展，如实时过程数据分析、云计算、连接和控制生产，使克服这些挑战成为可能。计划中的项目结合了人工智能（AI）、数值建模和过程技术领域的发展，以实现智能EMF过程的调查。基于数据的人工智能方法应确保系统的流程设计和控制。为此，从模拟和领域知识中获得的合成数据与传统的过程数据一起以一种技术上几乎没有探索的方法进行咨询。此外，还提出了一种基于综合数值模拟的电动势耐用工具设计方法。最后，将所有结果进行组合，并基于演示过程进行测试。