background.jpg

ODYSSEE

Optimal Decision Support System for Engineering and Expertise

OVERVIEW

ODYSSEE, is an innovative and powerful platform, allowing to process data from multiple horizons and build customized specific skill tools. Various features around machine learning, data handling, signal treatment, image processing & recognition, data compression and fusion are available in solver Quasar module. For Robustness analyses we propose to use optimizer Nova coupling with Quasar using specific Entropy indicators.

 

Main applications:

Why ODYSSEE ?

Our solution, ODYSSEE, helps our clients to reach the following strategic challenges:

Reduce Time

Simulations and optimization in real time contribute to Reduce design time

Reduce cost

Users can easily investigate and optimize their design space in real time and consequently “Reduce cost and delay” in analysis times as well as computational effort.

Protect the planet

Our customer can predict any simulation based on any physics in real time (solver independent), using simply a DOE of few simulations. We are very excited to contribute to the planet protection and customer cost-effectiveness.

 

Real-Time predictive modeling and optimization

Real-Time predictive modeling and optimization

Results prediction

simulations / experiments

Reduce computing effort

  • Few with wisely selected sampling points

  • Adaptive learning (improve as you learn)

Real-time computing

  • Zero-computing effort for parametric studies and optimization

Corridor / Population generation

Precision & completeness

  • Full time-history output (not only scalars)

  • Physical domain decomposition and not fitting (NOT a Response Surface Method!)

Solver & physics independent

Can produce 3D animations

  • No interpolations but reconstructions

  • Stress/displacement iso values reconstruction

Results optimization

simulations / experiments

Parametric sensitivity

Population/Corridors

Robustness

Optimization multi criteria, multi curves

Reverse engineering

Automation Methodology

simulations / experiments

Intelligent DOE

  • Adapted DOE tool

  • Possibility to improve an existing one

  • DOE can include simulations, tests or the both

Automation/Parser

  • To automatize the pre and post processing, the calculation and the results organization (LS-DYNA, MSC NASTRAN, MSC Adams or other with generic parser)

Methodology tools included

  • Quality of parameters,

  • Quality of DOE,

  • Best method for your application

Image compression, identification, learning, prediction

 

Image compression

Image

MMT Project Courtesy of Thales

Intelligent data compression on images
(best possible compression rate), while keeping as much as possible the useful information

Different methods have been integrated into Quasar

  • Images compression

  • Selective compression

Prediction and  Optimization for Designers 

Image

Machine learning based on CAD, associated parameters and results 

Prediction and Optimization for Health

Image (MRI files)

Machine learning based on MRI files, associated parameters and results 

Fault prediction

 

Fault characterization for quality control

Image 

Example: electronics - chip
(image analysis)​

Customer value: 

  • Identify defaults via image recognition

  • Save time via indicators for default types in real time

  • Reduced cost by early rejection of bad products

Methodology:

  • Learning from past images

  • Detecting defaults from new images (per chip)

  • Improved quality via Sorting and packaging in real time

Automation and quality

Simulation

Example: Automated optimal bolt tightening

Customer value: 

  • Save time via automatic calibration

  • Decision support tool for optimal calibration

Methodology:

  • Fast simplified model developed using FE computations

  • Technical know-how integrated in models

  • Tightening torque given for each cylinder

Fault prediction

Sensors

Example: Casting & Health monitoring

Customer value: 

  • Analysis of acquired on-board data

  • Cost saving by finding links between process parameters and defects

Methodology:

  • Data Analysis using physical sensor data

  • Correlation table and multi-plots using ODYSSEE.

  • Vertical application to prevent defects before they appear

©2020 CADLM. Created with Wix.com