Physics-Informed AI

Description

• Course ID:
  • IC-805-E-11
• Content: This course covers the use of artificial intelligence in combination with physical models and governing laws. The objective is to integrate data-driven AI models with physical knowledge in order to achieve better generalization, higher data efficiency, and more robust predictions in technical systems. The focus is on industrial, engineering, and simulation-based applications.
  • Key Topics:
    • Introduction to Physics-Informed AI (PIA) and hybrid modeling approaches
    • Motivation: limitations of purely data-driven AI in technical systems
    • Combination of physical equations and machine learning
    • Overview of Physics-Informed Neural Networks (PINNs)
    • Integration of differential equations into learning processes (conceptual understanding)
    • Modeling of physical systems with AI support
    • Data-scarce scenarios and the benefit of physical constraints
    • Simulation vs. data-driven modeling
    • Application areas in: - Fluid dynamics (CFD) - Thermodynamics - Structural mechanics - Electromagnetic systems
    • Validation of physics-informed models
    • Error analysis and interpretability of hybrid models
    • Limitations and challenges (computational cost, model stability)
  • Practical Components:
    • Development of a simple physics-informed model concept
    • Comparison of classical ML models vs. physics-informed approaches
    • Analysis of a technical simulation problem with AI support
    • Demonstration of a simplified PINN approach (tool- or notebook-based)
• Prerequisites:
  • Required:
    • IC-805-G-04 – Developing and Integrating AI Systems, or
    • Comparable technical AI and data science knowledge
  • Strongly recommended:
    • IC-805-G-10 – AI for Time-Series Data (for data-driven modeling in technical systems)
    • Basic knowledge of physics or engineering (depending on target audience)
    • Not suitable for complete beginners without technical or mathematical background.
• Certification Requirement:
  • Without certification option
• Conduction Method:
  • Online (For on-site or in-house training, please send us a request via our contact form (Contact - intellcert). We will get in touch with you shortly thereafter.)
• Language:
  • English
• Target  Audience:
  • Data scientists with technical or scientific focus
  • Engineers (mechanical engineering, electrical engineering, physics, simulation)
  • AI developers in industrial or research environments
  • Simulation engineers and computational scientists
  • R&D teams in industrial companies
  • Participants of IC-805-G-04 and IC-805-G-10
• Learning Objectives:
  • Participants will be able to:
    • Explain the difference between data-driven and physics-informed AI
    • Identify use cases for Physics-Informed AI
    • Conceptually design hybrid modeling approaches
    • Evaluate the benefits of physical constraints in AI models
    • Classify typical application areas in engineering and industry
    • Understand limitations and challenges of PINN-based approaches
    • Analyze or evaluate simple physics-informed model concepts
• Note:
  • Displayed price excl. VAT