How Danfoss Drives uses AI and simulation for engineering efficiency

MathWorks provides AI-based workflows for both battery modeling and BMS design. We asked Danielle Chu - How does AI help enhance the capabilities of battery modeling solutions? AI techniques help engineers create reduced-order battery models when they have large amounts of data from high-fidelity simulations or experimental testing. These AI-based models can then be integrated into Simulink for system-level simulations, C-code generation, and hardware-in-the-loop (HIL) testing, explained Danielle. For an insightful conversation on Battery Modelling and Simulation Using Data and AI, tune in - https://lnkd.in/gzSgMFPF.

Engineers today have access to more powerful tools than ever before - for design, simulation, visualization and more. But with more power comes more complexity - engineering is just as difficult as ever. This is where AI can come in - by being an assistant to the engineer, helping them iterate more quickly than before - helping them navigate challenging problems and information-dense work. Watch as Nexus, our AI Agent, helps navigate an engineer through a complicated safety test using Finite Element Analysis - making the entire process nearly 50 times faster, without any loss of accuracy.

https://lnkd.in/eNyePAww The AMPEL360 combines breakthrough technologies in a Blended Wing Body configuration to achieve: 90% reduction in operational emissions (H₂ fuel cells) 30% improvement in aerodynamic efficiency (BWB design) Carbon negative operations through active CO₂ capture 25% reduction in maintenance costs via CAOS (Computer Aided Operations and Services) AI system. Explore deeply the architecture through the OPT-IN framework: Organization, Program, Technologies (on board systems), Infrastructures (off board systems and interfaces), Neural Networks.

Thought Precision and Phase Invariance Today I revisited one of my recent control analyses entirely in my head — no calculator, no code, just clean mental modeling. I derived the full scaling relationship between thrust, exhaust velocity, power efficiency, and variance symmetry for a 1 N electric propulsion system operating at 2.43 kW and 98.6% efficiency. The result closed to the decimal: ve=4.925km/s Isp=502 sI sp =502s m˙=2.03×10−4kg/s What’s more, the model held phase invariance — its performance remained unbiased under any time shift or noise perturbation. In other words, the solution was stable under transformation, a hallmark of true system equilibrium. Moments like this remind me that engineering isn’t only about tools; it’s about the mind’s capacity to hold and resolve complexity. Whether you’re in aerospace, AI systems, or quantum modeling, the ability to perceive invariance within flux defines the upper tier of precision thinking.

AI isn’t just changing the way engineering teams work, it’s redefining what’s possible. From predictive simulations to energy optimization and intelligent fault detection, mechanical and electrical engineering teams are using AI to boost speed, precision, and performance across projects. Discover how real companies are applying these strategies today, the measurable results they’re achieving, and how the right talent can accelerate your AI adoption. Read the full blog: https://bit.ly/43mw9bs #LoveWhatYouDo #AIPoweredEngineering #MechanicalEngineering #ElectricalEngineering #AIinEngineering

Everyone talks about “AI vision” like it’s one magic box. But every reliable vision system is built on 7 critical layers. If one fails, the whole thing collapses. Here’s what the Machine Vision Stack really looks like 👇 1️⃣ Optics – The lens defines what the camera can truly see. 2️⃣ Lighting – Controls contrast and repeatability. Poor light = poor data. 3️⃣ Sensor – The camera chip and interface. Determines resolution, speed, and data volume. 4️⃣ Processing – The compute layer (IPC or edge device) that runs algorithms in real time. 5️⃣ Algorithms – The vision logic: filters, segmentation, pattern matching, deep learning. 6️⃣ Integration – PLCs, robotics, conveyors — where digital vision meets physical action. 7️⃣ Data Layer – Insights, reports, traceability — where value compounds over time. When you treat “vision” as just a camera purchase, you miss six layers of performance potential. At ATEK, our systems work because we design across the entire stack — not just install cameras. What’s the most underestimated layer you see in vision deployments?

Companies spend millions on simulation software and computing power. But there's a growing cost few talk about: the time engineers spend managing, searching for, and recreating past data, amounting to over half of their time spent on non-value added tasks that could be freed up for innovation. Yes, AI will transform product development, but only if engineering teams first have clean, structured, and contextualized data. We're hearing from even the most sophisticated teams that they're are still trying to solve this gap. Next week I'm showing how leading teams use Rescale Data Intelligence, our newest product, to solve both challenges. Automating data workflows while building the foundation for AI-driven engineering. Join me Wednesday 11/12 at 8:30 AM PT. Link in comments. #hpc #data #simulation #cae #modeling #engineering #plm #spdm

Three years ago, a friend introduced me to a concept that completely changed the way I think about process modeling: Hybrid Models. In my view, hybrid models combine the best of two worlds to develop digital twins — the phenomenological or first-principles approach, which explains how a process should behave according to the laws of physics, transport phenomena, thermodynamics, and chemical reaction kinetics, and the data-driven approach, which captures how the process actually behaves in real operation. Together, they form a dynamic representation that not only simulates but also learns, adapts, and predicts performance under real conditions. I have always found first-principles models fascinating. However, in many cases, to make them solvable, we must rely on multiple assumptions that simplify the real world: ideal separations and mixing, no dead time, no data latency, constant efficiencies, no equipment wear, or perfectly tuned control loops. These assumptions make computation easier but can overlook key phenomena that affect real plant behavior. Of course, if we only used process data, we could develop a purely machine-learning digital twin. But here lies another challenge: not all variables are measured or stored in real time, and even when they are, sensors may be uncalibrated, drift over time, or provide unreliable readings. That’s why I believe hybrid digital twins are both the present and the future, because they allow us to take advantage of the strengths of both approaches: the interpretability and structure of physics, and the adaptability and learning capacity of data. But of course, this new paradigm also brings new challenges: ensuring data quality and calibration, maintaining synchronization between physical and digital layers, and developing methods that remain explainable, scalable, and trustworthy as systems evolve. 💭 I’d love to hear your thoughts: Have you worked with hybrid models or any digital twin before? Did you already know about this concept? What challenges have you faced when combining phenomenological models and data-driven approaches? Let’s share ideas — I’d love to learn from your experience too.

Dassault Systèmes SIMULIA is embedding Artificial Intelligence and Machine Learning directly into our solutions, creating Generative Experiences that democratize simulation and accelerate design optimization. The core benefit? Time. ML models can deliver reliable simulation results in seconds, powering the MODSIM approach which unifies modeling and simulation. This is the key to faster product launches: "Dramatically reduce development time by up to 90% and from months to weeks." – Dassault Systèmes SIMULIA This massive acceleration—driven by smarter, AI-enabled workflows—allows engineers to explore a far greater design space, confidently reaching the optimal product faster than ever before. Learn more about how MODSIM and AI are accelerating product development: MODSIM Unified Modeling & Simulation https://lnkd.in/g4YY6ZMr #SIMULIA #MachineLearning #GenerativeDesign #MODSIM #3DEXPERIENCE #ProductDevelopment

Global automotive and industrial supplier Schaeffler is harnessing generative AI to revolutionize design, engineering, and manufacturing. By integrating Siemens Industrial Copilot with the Totally Integrated Automation (TIA) Portal, Schaeffler is transforming how teams work. How Schaeffler is leading the way: • Generate PLC code faster with natural language inputs • Identify and fix errors quickly to reduce downtime • Empower less-experienced employees with guided processes • Optimize machine performance from planning to operation “This is the beginning of a new era. In the past, we had to speak to machines in their language. With the Siemens Industrial Copilot, we can speak to machines in our language,” says Cedrik Neike, CEO of Siemens Digital Industries. https://sie.ag/5EDend https://sie.ag/St5MT