Products

Simcenter Studio 2410 – Next-generation AI/ML and SysML v2 at your fingertips

In a highly competitive market, standing still is as good as going backwards. What do you need to take your product design and engineering to the next level? How about:

  • Better collaboration, with seamless model exchange and reuse across engineering disciplines and expertise
  • Fully leveraging accelerating technologies like AI and ML
  • Immersion in digital thread initiatives like MBSE with SysML v2 to drive the ‘big picture’ of product performance

It feels like the world of computational engineering is braced for some disruptive change. The winners will be those who successfully jump onto these new vehicles for innovation. But how?

Meet Simcenter Studio

It’s both your advanced technology and data science playground and generative engineering workhorse. Simcenter Studio lets you consume, execute and analyze SysML v2.0 models with great efficiency, in alignment with the latest modeling standards and plugging you into the core of MBSE.

From there, you can start innovating. Simcenter Studio’s patented AI technology can automatically explore novel design possibilities through architecture generation. It will even evaluate design candidates using a simulation service to run models seamlessly with Simcenter Amesim or FMU.

But how can you evaluate a design if the control algorithm hasn’t been developed yet? Simcenter Studio has an answer for that too – a powerful reinforcement learning workflow designed for non-experts to rapidly train and deploy RL agents to steer your product through those early steps.

What’s new in Simcenter Studio 2410?

Let’s take a look at how the state of the art has been advanced over the past 6 months. 2410 includes powerful developments in all three avenues of technology innovation listed above.

Collaboration, model exchange and reuse

Simcenter Studio helps you to keep your thoughts and methods in order with Jupyter notebooks. These ‘computational notebooks’ can include models, code, expressions, commands as well as visualization and commentary. A notebook becomes a self-contained recipe for a particular operation, process or study.

Computational notebook pipelines

In version 2410, an entire notebook can now be called as a function from another notebook. This is a simple sounding but powerful capability, enabling you to more readily share knowledge and capability from your library of saved notebooks between programs, people, departments or organizations.

Instant nbzip preview

And browsing those notebook libraries just became easier, too. The handy nbzip (notebook zip) files now contain an HTML preview which can be instantly checked as you look for the right notebook to select and share.

FMI 2.0 Model Exchange support

FMI 2.0 is a crucial enabling technology for democratization and bridging organizational boundaries. Simcenter Studio leverages FMI for architecture evaluation (using the Simulation Service) in generative engineering, as well as an export option from the reinforcement learning workflow.

In addition to the FMI 2.0 Co-Simulation standard, Simcenter Studio 2410 now also supports the Model Exchange FMU type, including for the export of Reinforcement Learning models, resulting in greater compatibility with the tools or processes you interface with.

Easier, faster and more robust AI/ML workflows

As well as being a natural fit for many data science, AI and ML applications, Simcenter Studio also offers some specific functionality for augmenting engineering workflows with these technologies.

Generative Engineering

AI-driven architecture generation in Simcenter Studio is a unique patented technology designed to ‘think out of the box’ for you. Given a system architecture and a set of governing principles and constraints, Simcenter Studio can exhaustively generate all topologically unique system architecture candidates.

Version 2410 includes some new tools to make this even smoother. First, an improved architecture generation interface, with stop and resume controls and live feedback on solution progress. You also now have the freedom to start perusing the architectures generated right away and to evaluate them while the generation process continues in the background. Combined with the neat ability to include images on your block diagram, there is a lot to be enjoyed by generative engineers in this release.

Reinforcement Learning (RL)

One of the common challenges in leveraging simulation in the early stages of a design process is the need to understand performance over a drive cycle, or mission profile. In other words, going beyond single operating points to dig into the dynamic behaviour of a system. Doing this would require a controller algorithm to exist, which is generally not the case at the beginning of a design cycle.

Using Reinforcement Learning (RL) you can create a controller from scratch by allowing it to autonomously learn how to control your system by trial and error. Reinforcement Learning is a hugely powerful method, however in general it remains the domain of ML experts owing to the complexity of the discipline.

We are working to change that with Simcenter Studio, making RL a tool for all engineers, regardless of experience.

The reinforcement learning (RL) workflow in Simcenter Studio has seen some significant developments in recent releases. This continues in version 2410, with several more enhancements to the user experience.

These include the ability to name or define variables; making it easier to identify plant model inputs/outputs, easier validation of the environment and RL agent; with open-loop and stepwise closed-loop simulation execution, further flexibility in defining training and testing scenarios, and finally an improved control panel giving you a range of information on progress with pause/resume controls.

Simcenter Studio includes a broad range of state-of-the-art RL agents for different applications, as well as supporting the Gymnasium RL interface to connect to any other compatible agent not already included.

MBSE digital thread immersion powered by SysML v2

Left-shifting simulation, collaboration, democratization and model reuse all require all stakeholders to unite around a single digital thread. With everyone pulling in the same direction, you need to also be speaking the same language. And that language is surely SysML v2.

Simcenter Studio introduced one of the earliest implementations of SysML 2 in a production tool, and this has been growing from strength to strength in recent versions. In fact, Simcenter Studio now supports the complete SysML v2 language standard in its polylingual Jupyter notebooks.

The support for SysML v2 includes a programmatic interface, interactive help and demos, evaluation and validation of models against requirements as well as interactive visualisation of model trees.

A tool for the future engineer, today

Time and tide wait for none

English proverb

That’s right – there is no better time than now to start bringing these new technologies to bear on your engieering challenges. As you can see from the raft of new capabilities to accelerate and facilitate your innovation, Simcenter Studio is working hard to democratize these emerging and disruptive technologies and put them in the hands of every engineer.

When are YOU going to start using them to your advantage?

Alex Graham

Senior Marketing Manager, Simcenter

More from this author

Leave a Reply

This article first appeared on the Siemens Digital Industries Software blog at https://blogs.sw.siemens.com/simcenter/simcenter-studio-2410-ai-ml-sysml-v2/