Explore new Frontiers of Innovation

Support

modeFRONTIER Response Surface Methods (RSM)

Course overview

This training course is designed to help modeFRONTIER users to expand their knowledge of interpolation techniques. This course offers the possibility to learn more about basic mathematics that stands behind interpolation algorithms and some details about the algorithm implementation. The attendee will learn how to create a large variety of global approximation techniques from simple polynomial interpolations up to neural networks and statistical models. One of the course topics will be the configuration RSM parameters to improve the accuracy, robustness and convergence speed of optimization algorithms when applied to mathematical and industrial problems. Some hints about the choice of the right interpolation algorithm for a particular data set will be given and comparative examples will explain way of working of different interpolators. Moreover the attendee will learn how to evaluate the quality of the interpolation by the use of the modeFRONTIER graphical tools. Finally the attendee will learn how to use these virtual models for global optimization in order to lead to better design in less time.

Prerequisites

Getting started with modeFRONTIER or modeFRONTIER Fundamentals are recommended to get the best from this course.

Audience

The primary audience for this course includes new modeFRONTIER user who wants to have a complete and detailed explanation of all modeFRONTIER interpolation algorithms and some practical advices about how to use them in real world problems.

Topics

  • Explore the design space and Importance of correct samplings
  • Statistical analysis
  • Response surface methodologies:
    - Polynomial interpolation
    - Parametric surfaces
    - K-nearest method
    - Kriging method and Gaussian Processes
    - Neural Networks
  • RSM quality evaluation
  • Over-fitting and Outliers
  • Virtual optimization using RSM

At the end of the course, the attendee will be able to

  • Sample the design space properly
  • Select the most suitable response surface method
  • Construct and evaluate response surface
  • Use RSM to speed up the optimization process