About DML-CZ | FAQ | Conditions of Use | Math Archives | Contact Us
  Previous |  Up |  Next

Article

Mlýnek, Jaroslav ; Srb, Radek ; Knobloch, Roman
The use of graphics card and nVidia CUDA architecture in the optimization of the heat radiation intensity. (English). In: Chleboun, J., Přikryl, P., Segeth, K., Šístek, J. and Vejchodský, T. (eds.): Programs and Algorithms of Numerical Mathematics. Proceedings of Seminar. Dolní Maxov, June 8-13, 2014. Institute of Mathematics AS CR, Prague, 2015. pp. 150-155
MSC: 65Y05, 65Y20, 68W10, 90C26, 90C90
Keywords:
optimization; acceleration of computation; graphics card; CUDA architecture; parallel programming; differential evolution algorithm
Summary:
The paper focuses on the acceleration of the computer optimization of heat radiation intensity on the mould surface. The mould is warmed up by infrared heaters positioned above the mould surface, and in this way artificial leathers in the automotive industry are produced (e.g. for car dashboards). The presented heating model allows us to specify the position of infrared heaters over the mould to obtain approximately even heat radiation intensity on the whole mould surface. In this way we can obtain the uniform material structure of artificial leather. The gradient methods are not suitable to optimize the position of heaters because the minimized function contains many local extremes. Therefore, we used an evolutionary algorithm, specifically the differential evolution algorithm. In this case the optimization procedure needs a lot of operations (especially when the mould volume is large and we use a large number of heaters). A substantial acceleration of the calculation can be achieved by parallel programming using a graphic card and nVidia CUDA architecture. The numerical calculations were performed by the Matlab code written by the authors and were run on a standard PC.
Partner of
EuDML logo