. . . "2011-01-01+01:00"^^ . . . . . . . . . "Artificial neural networks in multi-scale modelling of transport processes in heterogeneous materials"@en . . "2015-05-22+02:00"^^ . . . "2014-12-31+01:00"^^ . "\u0158e\u0161en\u00ED a v\u00FDsledky projektu byly do jist\u00E9 m\u00EDry ovlivn\u011Bny n\u00E1stupem na mate\u0159skou dovolenou v pr\u016Fb\u011Bhu \u0159e\u0161en\u00ED projektu. Uv\u00E1d\u011Bn\u00E9 publikovan\u00E9 v\u00FDstupy Jimp byly dolo\u017Eeny a\u017E na vy\u017E\u00E1d\u00E1n\u00ED. D\u00E1le jsou dolo\u017Eeny dva podan\u00E9 \u010Dl\u00E1nky Jimp nevstupuj\u00EDc\u00ED zat\u00EDm do RIV. Byly nav\u00E1z\u00E1ny zahrani\u010Dn\u00ED spolupr\u00E1ce. V\u00FDsledky projektu odpov\u00EDdaj\u00ED jeho n\u00E1vrhu."@cs . "2014-04-18+02:00"^^ . "GPP105/11/P370" . "0"^^ . "http://www.isvav.cz/projectDetail.do?rowId=GPP105/11/P370"^^ . "Solutions and results of the project are influenced by the starting her maternity leave during the project. Reported Jimp published papers were documented up on request. Two submitted articles Jimp not entering into the RIV yet are attached to the final report as ouputs of the project. The international cooperation has been established. The project results correspond to its proposal."@en . . "Proposed projectis focused on design and study of different applications of artificial neuralnetworks in multi-scale modelling of transport processes in heterogeneousmaterials.The last years witnessed a rapid development notonly in computer technologies, but also in experimental equipment, whichenables to investigate materials at very small scales. Nevertheless, most ofengineering materials are heterogeneous at small scale. Common approach to modellingsuch material systems is homogenization, which is based on computation of effectiveproperties of a periodic cell. Despite of a significant progress inhomogenization theories during last decades, the effective properties werederived analytically only for heterogeneous materials with linear behaviour.Generally, the effective properties can be obtained numerically, but such stepincludes a time-consuming simulation by a finite element method. Therefore, itis advantageous to employ a neural network as an approximation of a complex relationfor effective properties and hence, to substantially speed up the wholemulti-scale simulation."@en . "25"^^ . "25"^^ . "1"^^ . . "0"^^ . . "Vyu\u017Eit\u00ED um\u011Bl\u00FDch neuronov\u00FDch s\u00EDt\u00ED p\u0159i v\u00EDce\u00FArov\u0148ov\u00E9m modelov\u00E1n\u00ED transportn\u00EDch proces\u016F v heterogenn\u00EDch materi\u00E1lech" . . . . "Navrhovan\u00FD projekt je zam\u011B\u0159en na navr\u017Een\u00ED a prostudov\u00E1n\u00ED r\u016Fzn\u00E9ho uplatn\u011Bn\u00EDum\u011Bl\u00FDch neuronov\u00FDch s\u00EDt\u00ED p\u0159i v\u00EDce\u00FArov\u0148ov\u00E9m modelov\u00E1n\u00ED transportn\u00EDch proces\u016F vheterogenn\u00EDch materi\u00E1lech.V posledn\u00EDch letech se vedle v\u00FDpo\u010Detn\u00ED technikyrozv\u00EDjej\u00ED tak\u00E9 experiment\u00E1ln\u00ED za\u0159\u00EDzen\u00ED, kter\u00E1 umo\u017E\u0148uj\u00ED zkoumat vlastnostimateri\u00E1l\u016F ve velmi mal\u00E9m m\u011B\u0159\u00EDtku. Od ur\u010Dit\u00E9ho p\u0159ibl\u00ED\u017Een\u00ED m\u00E1 ov\u0161em v\u011Bt\u0161ina in\u017Een\u00FDrsk\u00FDchmateri\u00E1l\u016F heterogenn\u00ED strukturu. Velmi roz\u0161\u00ED\u0159enou metodou modelov\u00E1n\u00ED takov\u00FDchsyst\u00E9m\u016F je homogenizace, kter\u00E1 je zalo\u017Een\u00E1 na v\u00FDpo\u010Dtu efektivn\u00EDch vlastnost\u00EDperiodick\u00E9 bu\u0148ky. I p\u0159es zna\u010Dn\u00FD pokrok v homogeniza\u010Dn\u00EDch teori\u00EDch v posledn\u00EDchdesetilet\u00EDch byly tyto vlastnosti odvozeny analyticky pouze pro velmi jednoduch\u00E9heterogenn\u00ED materi\u00E1ly, kter\u00E9 nav\u00EDc vykazuj\u00ED line\u00E1rn\u00ED chov\u00E1n\u00ED. Obecn\u011B lze tytovlastnosti ur\u010Dit numericky, to ale zahrnuje n\u00E1ro\u010Dn\u00FD kone\u010Dn\u011Bprvkov\u00FD v\u00FDpo\u010Det.Proto je zde v\u00FDhodn\u00E9 vyu\u017E\u00EDt neuronov\u00E9 s\u00EDt\u011B k aproximaci komplexn\u00EDch vztah\u016F proefektivn\u00ED vlastnosti a dos\u00E1hnout tak v\u00FDrazn\u00E9ho zrychlen\u00ED cel\u00E9 v\u00EDce\u00FArov\u0148ov\u00E9simulace heterogenn\u00EDho materi\u00E1lu." . "artificial neural network multi-scale modelling transport processes heterogeneous materials effective properties"@en .