Fakulteta za strojništvo, Ljubljana (FSLJ)
  • Multiple layer perceptron simulation and advanced metamodelling of centreless grinding process [Elektronski vir]
    Krajnik, Peter ...
    The enhancement of the centreless grinding process efficiency and understanding can be achieved by process simulation. In order to apply the simulation, a multiple layer perceptron artificial neural ... network (MLPANN) model is used to take on the role of the real plunge centreless grinding process. The experimental data were transferred into a simulation model that was used in predictive simulation of surface roughness within the region of experimental interest. The system based methodology treats the simulation model as a black box and requires the development of a metamodel. The metamodelling includes the design of experiments (DOE), in which a simulation input set is selected, predictive output simulation, and metamodel fitting. The developed metamodel is an exact interpolator, based on a Gaussian process regression (GPR), which is founded on the Latin hypercube sampling (LHS) of experiments.
    Vir: AMPT 2006, Las Vegas, NV, July 30-August 3, 2006 [Elektronski vir] (4 str.)
    Vrsta gradiva - prispevek na konferenci
    Leto - 2006
    Jezik - angleški
    COBISS.SI-ID - 9560859

Vnos na polico