As a pioneer in implementing AI/ML/ROM technology in engineering computer aided design (CAE), CADLM, editor of ODYSSEE platform, is proud to announce the new release of ODYSSEE.Lunar (4.1.0) and the new release of ODYSSEE.Quasar (2.1). ODYSSEE.Lunar is the ODYSSEE package dedicated to Machine Learning and Reduced Order Modeling (ROM. ODYSSEE.Quasar is the AI solver of ODYSSEE while ODYSSEE.Nova (2.1) is the stand-alone optimization package.
ODYSSEE, is an innovative and powerful platform, allowing to process data from multiple horizons and build customer specific tools easily. Various features such as machine learning, data handling, signal and image processing & recognition, data compression and fusion are available in Quasar module. For Robustness analyses we also propose to use optimizer Nova, coupled with Quasar, using our unique and innovative System Entropy based indicators.
The main applications of ODYSSEE are:
Real-Time predictive modeling and optimization
Image based compression, classification, learning, prediction
Fault prediction (sensor data)
ODYSSEE.Lunar uses Machine Learning and Reduced Order Modelling (ROM) techniques based on algebraic or fusion solutions for reducing the volume of data while preserving the most important parts of the information contained within that data. Such techniques allow for creating on-board and real-time applications based on existing experimental or simulation results (ex. finite elements (FEM)). Typical applications are optimization, parametric sensitivity analysis and robustness.
In addition to the previous version’s already available options such as DOE generation, prediction of global time responses, parametric studies, sensitivity analysis, optimization and reconstruction of the animation snapshots in real time, ODYSSEE.Lunar 4.1 now proposes an intelligent DOE generation (adaptive DOE tool, possibility to improve an existing design manually or automatically with detection of poorly sampled zones and automatic improvement). Additionally, Lunar can now manage, in a unique project, multiple output as well as optimization of multi constraints, multi-objectives, in real time.
Many available Quasar scripts support our customers automation needs in order to evaluate the quality of parameters and of the DOE, and to determine the best algorithm for a given application (and per output). ODYSSEE.Lunar can predict and optimize results for any physics (crash, statics, dynamics, acoustics, fluid, mold injection or other process, thermal, acoustics, aerodynamics, multi-bodies dynamics, … Many examples are available for different applications which may also be shared with customers.
“Our pioneer solution, ODYSSEE.Lunar, helps our customers to reach the following strategic challenges:
- Simulations and optimization in real time which contributes to reduction of design cost, including for time dependent physics,
- Investigation of digital twin concept via evaluating and visualizing the impact of change or uncertainty on the final product”
said Dr Ing. Kambiz Kayvantash, CTO and CEO of CADLM.
ODYSSEE.Quasar, the AI solver of ODYSSEE, is a very powerful tool and can provide readily available solutions for various techniques of Data Mining, process discovery, data fusion, pattern recognition, automation, machine learning and model reduction, Image or CAD compression, identification, learning, prediction,… all which can be managed by available user friendly GUI’s, simple scripting or programing or via encapsulation into other software (external software, or custom application for design, manufacturing , production, validation and verification … ) based on very low memory and CPU. The CADLM application library can help you to reach very efficient algorithms without extensive programming and can be connected to your own software library (via an Embedded memory sharing API). The coupling with Nova, our optimization tool, can propose a very efficient solution spending less than 10% of the simulation effort, usually required by traditional response surface methods.
“Our choice to develop a dedicated language in Quasar has been strategic and has proved to be very efficient for the development of proprietary algorithms and vertical applications in AI for different objectives and markets without any dependence on low reliability and low confidentiality web-based solutions such as:
- big data and machine learning: Real time analysis of large data bases via web application or support for diagnostics in healthcare using images, simulations, expertise
- CAE applications: Real-Time predictive modeling and optimization of time dependent processes using a few simulations or measurements with customized applications for designers,
- Health monitoring: Data & Images applications, data Fusion to predict stochastics scenario for autonomous vehicles, image-based predictions,
- Manufacturing: with Application in machining, injection, 3d printing …, etc.”
- Kambiz Kayvantash, CEO and CTO, CADLM