JuliaSIM

JuliaSim is a cloud-based platform developed by JuliaHub that aims to facilitate model-based design by combining scientific machine learning (SciML) techniques with equation-based digital twin modeling and simulation. According to the vendor, JuliaSim can significantly accelerate simulation, with claims of up to 500x faster speeds. It is targeted at engineers, scientists, and industrial researchers across small, medium, and large enterprises. The automotive and aerospace industries are specifically highlighted as potential beneficiaries of this software.

Key Features

Build Models Faster: JuliaSim is built on the ModelingToolkit acausal modeling platform, which, according to the vendor, allows engineers to create complex models with less code and greater code reuse compared to causal modeling platforms. Additionally, engineers can leverage pre-made models provided by JuliaSim, covering a range of domains from two-phase flow HVAC simulation to battery packs.

Transform Models to Digital Twins: JuliaSim provides tools to integrate data into the predictive modeling process. The vendor claims that through nonlinear calibration, integrated automatic differentiation, and optimal design tools, engineers can achieve fast and accurate parameter tuning. Furthermore, JuliaSim offers techniques like universal differential equations to automate the discovery of missing higher-order physics, enabling the extension of models in a data-driven manner.

Ensure a Robust Deployment: JuliaSim emphasizes safety in deploying nonlinear model-predictive controllers and integrating machine learning models into real-world processes. The platform offers tools for tuning nonlinear controllers using robust and uncertainty-aware techniques. Additionally, JuliaSim enables the analysis of uncertainty and errors in neural surrogate models, as well as the quantification of parametric uncertainty in model fits.

Integrate with Existing Tools: JuliaSim seamlessly integrates with the industry-standard Functional Mock-up Interface (FMI) used across various modeling tools. Engineers can generate a Functional Mock-up Unit (FMU) from their existing modeling tools, such as Simulink or Dymola, and import it into JuliaSim. For teams not using FMUs, function registration allows embedding calls to Julia, R, and Python models from within model blocks.

Precision Modeling With Digital Twins Means Faster Production and Lower Costs: JuliaSim's unified modeling frameworks enable the integration of machine learning with traditional simulation, addressing the challenge of building predictive models. The platform provides built-in libraries and machine learning techniques to automatically discover higher-order physics and calibrate digital twins to real-world data. According to the vendor, neural surrogates are used to accelerate simulations while generating highly accurate nonlinear controls for deployment on embedded devices.

JuliaSim is Simple: By utilizing acausal modeling and pre-made models, scientists and engineers can reportedly build realistic models in minutes, as per the vendor's claims.

JuliaSim is Fast: The vendor states that JuliaSim offers simulation speeds that are 10x to 1000x faster than other traditional products. Additionally, the platform seamlessly works on GPUs, further enhancing simulation performance.

JuliaSim Integrates Machine Learning: JuliaSim leverages scientific machine learning (SciML) methods to automate the discovery of missing physics, accelerate simulation with neural surrogates, and improve processes with deep-learned controllers, according to the vendor.