Authors: M Gobbi F Levi G Mastinu G Previati
Publish Date: 2014/08/21
Volume: 51, Issue: 3, Pages: 645-657
Abstract
The Fritz John conditions for Paretooptimality have been set in matrix form and used for introducing a procedure for the analytical derivation of the Paretooptimal set in the design variables domain Subsequently the derivation of the Paretooptimal set in the objective functions domain can be obtained if possible by a proper analytical derivation Both the objective and constraint functions are assumed to be available in analytical form and twice differentiable and convex or pseudoconvex The proposed procedure to find the Paretooptimal set is relatively simple The computation of the determinant of a matrix is required A symbolic manipulator can be exploited If there are two design variables and two objective functions the Paretooptimal set can be easily computed by applying a simple formula derived in the paper If the number of design variables equals the number of objective functions the Paretooptimal set in the design variables domain can be found by computing the product of the constraint functions times the determinant of the Jacobian of the objective functions A number of case studies have been proposed to test the effectiveness of the proposed procedure The optimal structural design of respectively a pair of compressed spheres a cantilever with rectangular cross section have been faced and solved Additionally the test problem proposed by Fonseca and Fleming has been addressed and solved analytically Optimization problems with low dimensionality 2 or 3 design variables and 2 objective functions 2 or more constraints have been easily solved The proposed procedure can be useful in the actual engineering practice at the earliest design stage In this case the designer can be made aware on the proper design variables setting to obtain the desired tradeoff among conflicting objective functions
Keywords: