Project Description

Design of Experiments and optimization of aircraft design

606px-Dassault_Rafale

In the frame of a collaborative project with Dassault Aviation, Scilab Enterprises developed an advanced modeling and optimization platform (Scilab Optimisation Platform) integrating features in data analysis and Design of Experiment (DoE), and covering the following functional fields:

  • Data management
  • Modeling
  • Optimization

Data management

  • Load and generate existing DOE (iSight,…) : possibility to add its own DOE generator
  • Response simulation using external tools (openFOAM, CATIA, CCM+,…) or Scilab functions
  • 2D visualization of factors and responses

Modeling

Selection among various modelers: DACE (Kriging), LOLIMOT (LOcal LInear Model Tree, a fast neural network) ,…

Parameter configuration

Multiple model management with best model selection

Possibility to select points:

  • Learning point
  • Validation points
  • Bad points (simulation issues,…)

Visualisation and optimisation

Execution and 2D visualization:

  • Response: all factors, two factors
  • Correlation

Responses coefficients setting

Optimizer:

  • Selection between various algorithms: non-linear (with optim), constraint (with fmincon), genetic algorithms (with optim_nsga2),…
  • Possibility to add its own algorithm
  • Interactive configuration

Visualization:

  • Optimal point
  • Pareto frontier
  • Robustness

 

Credits:

Photo: Airwolfhound / CC BY-SA
Kaboldy

Content: Read further on the ScilabTec presentation of Dassault