Flexible implementation of approximation concepts in an MDO framework

Abstract

This paper describes the flexible implementation of approximation concepts in an MDO framework offered by the commercial-off-theshelf software package iSIGHT. Three different types of approximation models-Response Surface Modeling, Taylor Series Approximations, and Variable Complexity Modeling-are implemented in such a way that they may be used interchangeably in any combination to approximate any segment of an MDO analysis problem, as well as the complete system analysis sequence as a whole. The novelty of the implementation is in the fact that all approximation models are implemented as "smart" objects. Each of the objects possesses certain characteristics defined by the "tuning parameters" specific to each approximation type, and a suite of methods (actions) which operate on the data of each model. The concept of Response Surface Modeling is taken to the extreme of simplicity and efficiency by using a minimum number of design analyses for model construction and then gradually improving the quality of the model following the path of the optimizer The category of Taylor Series Approximations actually spans a whole suite of approximation techniques that are based on using the first derivatives of the response function. Applications demonstrate the utility and flexibility of the approach.

Multidisciplinary Environments: A History of Engineering Framework Development

“…By using inheritance, each new optimization routine must only supply code for those functions that are unique to that method. In a 1998 paper, Engineous Software, Inc. announced the iSIGHT framework, the first commercial framework to incorporate an object-oriented optimization scheme 8 . The iSIGHT framework treated optimization, approximation and probabilistic analysis routines as interchangeable objects associated with data and with methods.…”

“…On the other hand, the developers of MDICE prefe r a scripting language for execution control and a GUI to monitor progress and to associate codes with remote host machines 22 . As a third example, the developers of iSIGHT provide a GUI to monitor progress, to choose optimization methods and to build the scripting language 8 . In 2000, the FIPER team put some order into the discussion of user interface 41 .…”

A Systematic Optimization Design Method for Complex Mechatronic Products Design and Development

“…Wang and Shan [12] described the meta-modelling techniques in support of engineering design optimization. Golovidov and Kodiyalam [13] described the ideas and methods of how to use an approximate model to do multidiscipline optimizations. Jiang et al [14] used a neural network model to realize a multiobjective optimization involving process parameters.…”

“…Some papers suggest that the surrogate modelling technique can be applied to this problem [12][13][14][15]. When focusing on high-speed train, we choose the typical neural network surrogate model based on design space reduction and intelligent optimization algorithm to optimize its design.…”

Decentralized Design Under Uncertainty: Investigating the Impact of Designer Mistakes

“…The multiobjective optimization problem comprises of the local objective and the approximated non local objective. Using metamodeling techniques [41][42][43][44][45][46][47] , it is possible to develop an approximation for the non local objective function from the design vectors being communicated between the various subsystems. The communicated design variable vectors are used to model the rational reaction set which are in turn used to approximate the non local objective functions.…”