Efficient Automatic Vehicle Shape Determination using Neural Networks and Evolutionary Optimization

Visa enkel post
dc.contributor.authorLundberg, Anton
dc.contributor.departmentChalmers tekniska högskola / Institutionen för tillämpad mekaniksv
dc.contributor.departmentChalmers University of Technology / Department of Applied Mechanicsen
dc.date.accessioned2019-07-03T13:32:37Z
dc.date.available2019-07-03T13:32:37Z
dc.date.issued2014
dc.description.abstractRecent years have seen decreasing emission limits for passenger cars put in place to battle climate change. There is a need for car manufacturers to apply state-of-the-art techniques in order to be able to further reduce emissions and meet these new limits. Improving the aerodynamic shape of a vehicle still holds a large potential for cuts in emissions. A fast method for vehicle shape optimization have been developed using recent years' advancements in neural networks and evolutionary optimization. It requires the construction of morphing boxes as the only manual work, with everything else being automated. The proposed method enables a study of several design parameters to be carried out in a short period of time. This is great improvement over a classical approach of changing one parameter at a time. The optimization method is a type of two-level optimization. This means that the optimization is performed on a solver approximation instead of the real solver. This considerably reduces computation time. First a database is generated from simulations on a number of vehicle shape configurations. The configurations are chosen using a latin hypercube sampling where the minimum distance between any two points is maximized. The database is used to train a neural network to act as an approximation to the simulations. Finally an optimal vehicle shape is determined from the neural network using particle swarm optimization. The method can handle multiple objectives. The method was incorporated in an optimization tool compatible with Volvo Car Group's CAE process. The optimization tool was used on a simplified low-drag car model in a study of realistic changes of five design parameters. An improved shape with a 12.6% lower drag coefficient (CD) was achieved. The prediction error of CD was 0.3%.
dc.identifier.urihttps://hdl.handle.net/20.500.12380/204825
dc.language.isoeng
dc.relation.ispartofseriesDiploma work - Department of Applied Mechanics, Chalmers University of Technology, Göteborg, Sweden : 2014:18
dc.setspec.uppsokTechnology
dc.subjectStrömningsmekanik och akustik
dc.subjectHållbar utveckling
dc.subjectEnergi
dc.subjectTransport
dc.subjectFluid Mechanics and Acoustics
dc.subjectSustainable Development
dc.subjectEnergy
dc.subjectTransport
dc.titleEfficient Automatic Vehicle Shape Determination using Neural Networks and Evolutionary Optimization
dc.type.degreeExamensarbete för masterexamensv
dc.type.degreeMaster Thesisen
dc.type.uppsokH
local.programmeApplied mechanics (MPAME), MSc
Ladda ner
Original bundle
Visar 1 - 1 av 1
Bild (thumbnail)
Namn:
204825.pdf
Storlek:
4.52 MB
Format:
Adobe Portable Document Format
Beskrivning:
Fulltext
Ladda ner
Samling
Examensarbeten för masterexamen