This paper presents a case study involving a heavy truck side bumper and the optimal design concerning its vibrating aspects. The dynamical behaviour is modeled by means of response surfaces constructed from the results of computer experiments. These statistical meta models, presented as closed form polynomials are then used within different optimization approaches: gradient optimization techniques, such as the modified feasible directions method, in comparison with intelligent optimization methods, such as genetic algorithms, simulated annealing, and tabu search. Besides comparing the optimization results obtained from the two basic approaches mentioned above, the robustness of the intelligent algorithms is verified by varying the random sets of initial designs. It is important to mention that intelligent techniques are difficult to be used in many design applications because of the high computational cost to compute the fitness function. This computational overhead can be significantly reduced when using meta models to represent the fitness function. In this paper this is achieved by creating response surfaces. Numerical results support these assumptions.
[1]
Geoffrey E. Hinton,et al.
Learning representations by back-propagating errors
,
1986,
Nature.
[2]
George E. P. Box,et al.
Empirical Model‐Building and Response Surfaces
,
1988
.
[3]
Duc Truong Pham,et al.
Intelligent optimisation techniques
,
2000
.
[4]
David E. Goldberg,et al.
Genetic Algorithms in Search Optimization and Machine Learning
,
1988
.
[5]
Garret N. Vanderplaats,et al.
Numerical optimization techniques for engineering design
,
1999
.
[6]
Geoffrey E. Hinton,et al.
Learning representations of back-propagation errors
,
1986
.
[7]
Zbigniew Michalewicz,et al.
Genetic Algorithms + Data Structures = Evolution Programs
,
1996,
Springer Berlin Heidelberg.