ODYSSEE　機械学習を活用した設計・生産支援ツール　

CAEのデザインスペースの探索手法を得て、幅広い視点へ

ODYSSEE

Solving your engineering problems in real-time

Ready to get started?

Talk to an expert and discover how Hexagon can support your organisation.

製品概要

ODYSSEE CAE

ODYSSEE CAEはエンジニアが機械学習、人工知能、次元低減モデリング(ROM)および最適化設計をワークフローに適用できるようにするユニークで強力なCAEを中心とした画期的なプラットフォームです。

ODYSSEEの特長

機械学習と人工知能
統計、データマイニング、データ融合
最適化とロバスト性解析
プロセス探索
画像認識と圧縮処理

Lunarの機能

インテリジェントDOE

DOE(実験計画法）ツールを採用
既存のDOEツールを改善

リアルタイムコンピューティング

パラメータスタディと最適化にかかる計算負荷を削減
設計変数の変化範囲に対応する応答量の範囲予測

多くのソフトウェアと物理現象に対応

すべてのCAEソフトウェアのデータを利用可能
構造解析、熱解析、流体解析、音響解析に対応
CAEデータだけでなく、実験データも利用可能

自動化

プリ/ポスト処理の自動化
プリ：Nastran
ポスト：Marc、Nastran、Cradle CFD

解析計算負荷の削減

サンプリングポイントを厳選し、負荷を軽減
アダプティブラーニングシステムにより、学習と改善を並行処理

精度と完全性

フルタイムのヒストリーファイル出力
物理領域の分解を用いた単純な応答曲面法とは異なるモデル化手法

3Dアニメーション生成

内挿ではなく再構成
応力/変位の等値再構成

評価ツール

パラメータ品質評価
DOE品質評価
アプリケーションに最適な方法を探索

Nastran Sol 700とODYSSEEによる最適化

Nastran Sol 400とLunarを用いたリバースエンジニアリン

Nastran Sol 400とLunarを用いたリバースエンジニアリング

Marc-CRADLE連成解析をLunarで最適化

ベネフィット

リアルタイム・シミュレーション

わずかなシミュレーションに基づくパラメトリック設計と最適化により、プロジェクトの主要なステップをリアルタイムで管理できます
- 概念設計：パラメトリックスタディー、トライアンドエラー
- 詳細設計：最適化、モデルフィッティング
- 検証：信頼性・ロバスト性解析

ODYSSEEは社内のあらゆる関係者にベネフィットをもたらします

プロジェクトマネージャー
CAE解析部門
技術部門責任者
購買・調達部門

Challenges tackled with AI/ML

Inefficient processes –

Manual processes that are time-consuming and error-prone that lead to lower efficiency and productivity.

Higher costs –

Expensive manual processes that result in higher operational costs.

Limited scalability –

Difficulty scaling operations, handling large datasets and complex processes.

Inaccurate decision-making –

Limited data available to make informed decisions, leading to inaccurate or suboptimal outcomes.

Lower quality –

Lower quality products or services that lack capabilities to optimise processes and identify areas for improvement.

Overcome your challenges with ODYSSEE solutions

Reduce costs by optimising processes, predicting maintenance needs, and minimising waste

Better decision-making and faster predictive analysis & optimisation through AI/ML algorithms

Identify patterns and trends in data that allow faster predictions and forecasts

Adaptive learning with continuous improvement

Scalability to handle large and complex data sets and processes

ODYSSEE CAE

ODYSSEE CAE

Perform CAE design space exploration to broaden your horizons

ODYSSEE CAE is an innovative platform for engineers that integrates machine learning, artificial intelligence, reduced order modelling, and design optimisation into workflows. It enables cost-efficient digital twin creation through real-time predictive modelling and optimisation by leveraging CAE simulation and physical test data. With minimal prior simulations, ODYSSEE CAE can quickly generate accurate results (with deformation, stresses, ...) and full-time history outputs.

ODYSSEE A-EYE

ODYSSEE A-EYE

Image based learning and prediction

ODYSSEE A-Eye is a machine learning solution for accelerating product design and development through image and CAD-based learning. Using images, CAD and sensor data as inputs, ODYSSEE A-Eye creates customised AI applications that predict output responses and field data. This enables designers and production technicians to explore the design space thoroughly and design next-generation products efficiently without excessive computing time and cost.

ODYSSEE Solver

ODYSSEE Solver

The powerful AI engine

ODYSSEE Solver includes the Artificial Intelligence engine Quasar and the optimiser Nova. Quasar is an object-oriented programming language that allows configurable computations for data mining, prediction, interpolation, classification, image processing, matrix operations etc. Quasar is the AI engine powering ODYSSEE products, and it is integrated into multiple Hexagon products:

Cradle
MSC Nastran through smart sub-modeling
Adams through FMU

... and more ...

ODYSSEE Explore

ODYSSEE Explore

Data analysis and data mining managed by interactive dashboards

ODYSSEE Explore is an application for exploratory data analysis and data visualisation:

Perform data preprocessing and feature engineering before using the data for machine learning.
Visualise output responses and outliers.

ODYSSEE CAE can predict historical curves and complete field data at all nodes and elements (stress, strain, ...) in a way that is completely data-driven and CAE solver independent.

ODYSSEE can deploy different machine learning and reduced order modeling techniques to extract important features from your data. This enables the creation of highly accurate data-based surrogate models.

ODYSSEE ROM-based optimisers can perform efficient optimisation in real-time. Multi-objective and multi-constraint optimisation studies can be defined with ease.

To be more efficient during data preparation and preprocessing, ODYSSEE CAE proposes dedicated parsers (MSC Nastran, Adams and LS-DYNA) and a multi-purpose generic parser adaptable for any software.

The parsers dedicated for Marc, Cradle CFD, and MSC Apex are respectively integrated into Mentat ScFlow and MSC Apex.

Different algorithms are available for DOE sampling. These algorithms allow adding new sample points to improve space-filling. They also allow reducing the sample points to effectively handle large datasets.

ODYSSEE enables adaptive sampling that generates results in real-time when new points are added.

Parametric superelement (for MSC Nastran or Marc), parametric FMUs (for Adams and electronics devices...) can be generated with ease in ODYSSEE.

Replace a complete model with an ODYSSEE representative element for quick parametric studies and faster computation times.

Process images (JPG and PNG formats) and CAD files (STEP and STL formats) directly in A-Eye. Easily associate these with scalars, curves and labels.

Images can be geometry design views, manufactured parts snapshots or simulation field data.

The capability can be implemented for quality inspection, manufacturing fault detection and forecasting.

ODYSSEE Solver is built on an object-oriented language called Quasar. But it also supports Python.

Easy to automate any process or integrate into internal software.
Quasar is a powerful language, but engineers are familiar with Python. They can access ODYSSEE libraries with maintenance and support.
Built-in AI/ML, ROM, STATS, OPTIMIZATION, RELIABILITY, FORECASTING, and process monitoring libraries.
"No-code" access to all the above (very little scripting, just a couple of lines...).

"ODYSSEE CAE enables a comprehensive crash model to be run on a laptop with a run-time reduction of 95% compared to a research cluster. The reduced time and power needed allow parametric optimization for realistic crash simulation to improve safety and to set new trends in safety analysis."

Bengt Pipkorn

Director of simulation and active structures

Autoliv

"ODYSSEE CAE provided a straightforward solution to a very large optimisation study including time/history information."

Jesper Christensen

Associate Professor

Cranfield University

"Whereas finite element solutions would have taken more than 1000 hours to achieve the solution for the full design space, ODYSSEE CAE was able to accurately predict the full response within seconds."

Laike Misikir

Vehicle crash safety and CAE Engineer

Ford Motor Company

"ODYSSEE CAE helps Faurecia reduce the time required for crash safety simulations by over 97% for the Cockpit of the Future. Hard test results can be used to predict a different test configuration thanks to ROM."

Steeve Laine

Calculation/Simulations Engineer

Faurecia Brieres Tech Center

"Frequency dependent damper modeling with ODYSSEE helped us to explore beyond traditional modeling of damper characteristics in MBD simulations. This gave us the impetus to incorporate more complex damper characteristics and improve the fidelity of the simulation models."

Lenka Visweswara Rao

Senior Engineer (CAE Vehicle Dynamics)

Mahindra & Mahindra Ltd.

"Beyond simply increasing design efficiency, ODYSSEE CAE will be used to quickly quantify and evaluate digital data, thus enabling new designs without relying on a few individual skills."

Hiroshi Nasuno PhD

Team leader of Mechanism Analysis Group

Taica Corporation

"By using the ML (ODYSSEE), we can reduce the number of Finite Elements Method (FEM) by about 87% with high estimation accuracy."

Prof. Yuki Hidaka

Associate Professor

Nagaoka University of Technology

"By using ODYSSEE ROM, the results of parameter changes can be easily obtained by non-specialists without having to worry about complicated setting changes or errors."

Masahiro Takeda

CAE Engineer

JSOL

"Real-time computing because you literally spend no time doing any parametric studies, it only takes a few seconds."

Divesh Mittal

Engineering & Software Manager

ETA

"The approach in this study is unique, as it can capture the pertinent parameters influencing the casing buckling and the evaluation of the magnitude of each."

James Njuguna

Professor of Composite Materials

Robert Gordon University