Numerical Modelling and Optimization Including Stochastic Approach in Engineering Problems

Paper in proceedings (conference paper)

en

The purpose of the article is to introduce engineers to stochastic modelling approaches. In particular, stochastic optimization approach, which takes into account randomness and uncertainty of given parameters in a particular model, is described. Further, a need of their usage and resulting benefits in engineering problems are discussed. The advantages of that approach are listed and compared with disadvantages with respect to ordinary deterministic modelling approaches that include only fully-known parameters without embracing uncertainty and randomness although they are naturally comprised in real engineering problems. Unfortunatelly, the omitting of randomness in assembled models and their solutions may lead to incorrect results that do not correspond to real outcomes. To solve these stochastic optimization problems, the effective decomposition method for solving scenario-based stochatic optimization programmes – progressive hedging algorithm – and its usage with the possibility of parallelism are introduced. The outline of using parallel impementation of the progressive hedging algorithm and stochastic optimization techniques in the continuous casting control problem are described and also references to other applications of stochastic programming approach in metallurgy and engineering are included.

numerical modelling, stochastic optimization, progressive hedging algorithm, parallel implementation, continuous steel casting process

2011-04-06

TANGER, s.r.o.

Ostrava

978-80-87294-21-5

Sborník přednášek z konference Teorie a praxe výroby a zpracování oceli 2011

70–76

7