Artificial intelligence experts: PI Probaligence GmbH joins the CADFEM group
Artificial intelligence (AI) is fundamentally changing the way we live and work. Particularly in areas where the level of digitization is already high, AI is the key to a huge leap forward in efficiency; digital simulations and engineering processes are ready for implementation. CADFEM has enlisted PI Probaligence as a partner with outstanding solutions and expertise to provide its customers with targeted support as they enter the world of AI.
Machine learning in action: optimizing the adaptive model (Image: PI Probaligence GmbH)
The megatrend of artificial intelligence (AI) is on everyone’s lips today. Huge amounts of information are being processed in artificial neural networks using high-performance computers and cloud computing to interpret or generate text and images.
Other AI approaches specialize in design optimization, experimental planning or revealing important correlations. These include machine learning (ML), which is an approach to AI that learns by training on experiments, simulations and other engineering data. As a result, machine learning offers enormous potential for using an existing data set to massively accelerate and improve development and decision-making processes.
Machine learning (ML): AI for engineering
CADFEM, the simulation and digital engineering specialist, aims to help customers implement new technology in their digital process infrastructure so that its potential can be fully and sustainably utilized. This includes, for example
- Real-time applications for digital twins,
- Autonomous decision-making processes or
- Intuitive workflows that make it much easier to use AI for individual questions.
CADFEM has put in place the necessary technical expertise. These skills are complemented by close exchanges with ANSYS, Inc. which includes the new Ansys AI solutions in the CADFEM portfolio. CADFEM also relies on a tried-and-tested strategy: for major topics of the future, the company’s network is extended to include specialized partners and their expertise. This is done under the umbrella of the international CADFEM group, so that the additional know-how can be used worldwide.
The possibilities of AI for simulation are immense, for example for additional product optimization, digital twins or autonomous decision-making processes.
Probaligence: AI experts who understand simulation
When selecting a network partner for AI, PI Probaligence GmbH quickly emerged as the perfect option for CADFEM and the requirements identified by CADFEM’s customers. The main decisive factors were:
- PI’s clear focus on ML in combination with efficient stochastic methods
- The PI’s philosophy of closely integrating innovative research with practical relevance to industry is a perfect fit for CADFEM.
- IP STOCHOS’ unique algorithm, which combines expert independence, efficient use of IT resources and high-quality results.
Thanks to the combined approach of neural networks and Gaussian processes, STOCHOS offers outstanding predictive accuracy, even with a small database. STOCHOS is also characterized by its ability to reliably evaluate real and noisy data from a wide variety of sources, including measured values, simulation results and relevant key figures, and even to combine them in a meaningful and resource-efficient way.
Finally, PI’s proximity to simulation is another strong argument. On the one hand, PI already works with ANSYS, Inc. and, on the other, PI’s founders are very familiar with simulation methods due to their previous professional and scientific activities.
By acquiring a stake in PI Probaligence GmbH, the CADFEM Group has further expanded its family of solution partners, who combine important megatrends with the field of simulation within the global network.
The possibilities of AI for simulation are immense, for example for additional product optimization, digital twins or autonomous decision-making processes.
The heart of STOCHOS is the DIM-GP algorithm (DIM-GP = Deep Infinite Mixture of Gaussian Processes; this unique algorithm combines the advantages of two normally opposing AI approaches, neural networks and Gaussian processes. The resulting non-stationary probabilistic model is infinitely scalable – it no longer requires hyperparameter tuning and processes data of any dimension with minimal hardware requirements and without recourse to cloud computing. (Image: PI Probaligence GmbH)