This content is not included in your SAE MOBILUS subscription, or you are not logged in.
Windshield Shape Optimization Using Neural Network
ISSN: 0148-7191, e-ISSN: 2688-3627
Published March 08, 2004 by SAE International in United States
Annotation ability available
Wipe quality of wiper systems is influenced not only by the definition of the wiper system, but also by the shape of the glass. In order to optimize the overall performance of the system, Valeo Wiper Systems has developed an optimization algorithm, which is based on geometrical criteria. The multi-criteria objective not only considers wipe quality but also constraints by glass feasibility and respect of optical standards. As the direct derivation of the objective functions is not available, a neural network approximation is used at the place of the real function. A neural network with several outputs enables the engineer to include his knowledge in the optimization loop by changing disciplinary weights.
CitationMuradore, F., Dreher, T., and Jan, S., "Windshield Shape Optimization Using Neural Network," SAE Technical Paper 2004-01-1156, 2004, https://doi.org/10.4271/2004-01-1156.
- Grenouillat R., “Etude des défauts d'essuyage: application à l'analyse prédictive”, Ph.D. Thesis, Ecole Centrale de Lyon, 2001.
- Jallet S., et al., “Numerical Simulation of Wiper System Aerodynamic Behaviour”, SAE Article, 2001.
- Griewank A., “Evaluation Derivatives Principles and Techniques of Algorithmic Differentiation”, Society for Industrial and Applied Mathematics (SIAM), 2000.
- Van Grieken M., “Optimisation pour l'apprentissage et apprentissage pour l'optimisation”, Ph.D. Thesis, Université Paul Sabatier, to appear, (in French).
- Hornik K., Stinchcombe M., White H., “Multilayer Feedforward Networks are Universal Approximators”, Neural Networks, Vol. 2, pp. 359-366, 1989.