Project detail

Response surface and sensitivity analysis methods in stochastic computational mechanics (RESUS)

Duration: 01.01.2018 — 31.12.2020

Funding resources

Czech Science Foundation - Standardní projekty

- whole funder (2018-01-01 - 2020-12-31)

On the project

Navrhovaný projekt je zaměřen na systematické studium a rozvoj všech aspektů pokročilých metod plochy odezvy a citlivostní analýzy ve stochastické výpočtové mechanice. Hlavní motivací projektu je potřeba pracovat efektivně s časově náročnými výpočtovými modely na stochastické úrovni. Je třeba takových metod, které aproximují původní, výpočtově náročné modely jednoduššími, jejichž vyčíslení je rychlé. V rámci projektu budou vyvíjeny metody plochy odezvy založené na polynomické aproximaci a umělých neuronových sítích. Citlivostní analýza má těsný vztah k modelování metodami plochy odezvy, neboť pomocí citlivostní analýzy získáme vhled do chování výpočtového modelu. Citlivostní analýza modelu poskytne informaci o relativní významnosti každého vstupního parametru. Budou vyvinuty a aplikovány především tyto metody citlivostní analýzy: Metody založené na statistické simulaci, neparametrická pořadová korelace, citlivost pomocí souboru umělých neuronových sítí a citlivost založená na Sobolových indexech.

Description in English
The proposed project is focused on systematic treatment and progress of all aspects related to advanced response surface and sensitivity analysis methods in stochastic computational mechanics. The need to work effectively with time demanding computational models at stochastic level is the main motivation of the project. Methods in substituting the initial expensive model by a simpler model, fast to evaluate, are needed. The response surface methods will be developed based on polynomial approximation and artificial neural networks within the framework of the project. Sensitivity analysis has close relation to response surface modelling, as through sensitivity analysis we gain essential insight on computational model behavior. Sensitivity analysis of a model aims at quantifying the relative importance of each input parameter. Statistical simulation-based sensitivity analysis, non-parametric rank-order correlation, artificial neural networks ensemble-based parameter sensitivity analysis and sensitivity based on Sobol´s indeces will be mainly developed and applied.

Keywords
Metody plochy odezvy;citlivostní analýza;spolehlivost konstrukcí;umělé neuronové sítě;stochastická výpočtová mechanika;

Key words in English
Response surface methods, sensitivity analysis, structural reliability, artificial neural networks, stochastic computational mechanics

Mark

18-13212S

Default language

Czech

People responsible

Novák Drahomír, prof. Ing., DrSc. - principal person responsible

Units

Institute of Structural Mechanics
- (2017-03-28 - not assigned)

Results

NOVÁK, L.; NOVÁK, D.; SLOWIK, O. Application of polynomial chaos expansion to reliability analysis of prestressed concrete roof girders. In ENGINEERING MECHANICS 2018 PROCEEDINGS, VOL 24. First. Praha: Institute of Theretical and Applied Mechanics of the Czech Academy of Sciences, 2018. p. 609-612. ISBN: 978-80-86246-91-8.
Detail

LEHKÝ, D.; ŠOMODÍKOVÁ, M.; LIPOWCZAN, M.; NOVÁK, D. Inverse response surface method for prestressed concrete bridge design. In Bridge Maintenance, Safety, Management, Life-Cycle Sustainability and Innovations. 1. London: CRC Press/Balkema, 2021. p. 179-185. ISBN: 9780367232788.
Detail

LEHKÝ, D.; NOVÁK, D.; NOVÁK, L.; ŠOMODÍKOVÁ, M. Prestressed concrete roof girders: Part II – Surrogate modeling and sensitivity analysis. In Life Cycle Analysis and Assessment in Civil Engineering: Towards an Integrated Vision: Proceedings of the Sixth International Symposium on Life-Cycle Civil Engineering (IALCCE 2018). 1. London: CRC press, Taylor and Francis group, 2018. p. 495-502. ISBN: 9781138626331.
Detail

KALA, Z. Probabilistic verification of structural stability design procedures. Open Civil Engineering Journal, 2018, vol. 12, no. 1, p. 283-289. ISSN: 1874-1495.
Detail

SLOWIK, O.; LEHKÝ, D.; NOVÁK, D. Reliability-based design optimization using artificial neural network inverse analysis. 16th International Probabilistic Workshop. Beton- und Stahlbetonbau. 2018. p. 1-6. ISSN: 1437-1006.
Detail

PAN, L.; LEHKÝ, D.; NOVÁK, D.; CAO, M. Sensitivity analysis for parameters of prestressed concrete bridge using neural network ensemble. In Engineering Mechanics 2018. Engineering Mechanics .... Svratka, Česká republika: 2018. p. 637-640. ISBN: 978-80-86246-88-8. ISSN: 1805-8248.
Detail

ŠOMODÍKOVÁ, M.; LEHKÝ, D. An adaptive ANN-based inverse response surface method. Beton und Stahlbeton, 2018, vol. 113, no. S2, p. 38-41. ISSN: 0005-9900.
Detail

NOVÁK, L.; NOVÁK, D. Surrogate modelling in the stochastic analysis of concrete girders failing in shear. In Proceedings of the fib Symposium 2019: Concrete - Innovations in Materials, Design and Structures. International Federation for Structural Concrete, 2019. p. 1741-1747. ISBN: 9782940643004.
Detail

NOVÁK, L.; NOVÁK, D. On the possibility of utilizing Wiener-Hermite polynomial chaos expansion for global sensitivity analysis based on Cramér-von Mises Distance. In Proceedings of 2019 International Conference on Quality, Reliability, Risk, Maintenance, and Safety Engineering, QR2MSE 2019. 2019. p. 1-9. ISBN: 978-172-811-427-9.
Detail

NOVÁK, L.; NOVÁK, D. Moment independent sensitivity analysis utilizing polynomial chaos expansion. 2019. p. 1-6.
Detail

KALA, Z. Quantile-oriented global sensitivity analysis of design resistance. Journal of civil engineering and management, 2019, vol. 2019, no. 25(4), p. 297-305. ISSN: 1392-3730.
Detail

NOVÁK, L.; NOVÁK, D. Stochastic Spectral Methods in Uncertainty Quantification. Transactions of the VŠB – Technical University of Ostrava, Civil Engineering Series, 2019, vol. 19, no. 2, p. 48-53. ISSN: 1804-4824.
Detail

NOVÁK, L.; NOVÁK, D. On Taylor Series Expansion for Statistical Moments of Functions of Correlated Random Variables dagger. Symmetry, 2020, vol. 12, no. 8, p. 1-14. ISSN: 2073-8994.
Detail

PAN, L.; NOVÁK, L.; NOVÁK, D.; LEHKÝ, D.; CAO, M. Neural network ensemble-based sensitivity analysis in structural engineering: Comparison of selected methods and the influence of statistical correlation. COMPUTERS & STRUCTURES, 2021, vol. 242, no. 1, p. 1-19. ISSN: 0045-7949.
Detail

NOVÁK, L.; NOVÁK, D. On Taylor series expansion for statistical moments of functions of correlated random variables. In AIP Conference Proceedings. AIP conference proceedings. New York, USA: American Institute of Physics, 2020. p. 1-4. ISBN: 978-0-7354-4025-8. ISSN: 0094-243X.
Detail

SLOWIK, O.; LEHKÝ, D.; NOVÁK, D. Combinatorial reliability-based optimization of nonlinear finite element model using an artificial neural network-based approximation. In Lecture Notes in Computer Science. Siena, Italy: 2021. p. 359-371. ISBN: 978-3-030-64583-0.
Detail

PAN, L.; NOVÁK, D.; NOVÁK, L. Surrogate modelling of concrete girders using artificial neural network ensemble. In Life-Cycle Civil Engineering: Innovation, Theory and Practice. Proceedings of the 7th International Symposium on Life-Cycle Civil Engineering (IALCCE 2020), October 27-30, 2020, Shanghai, China. Shanghai, China: 2020. p. 1-5. ISBN: 9780429343292.
Detail

NOVÁK, L.; NOVÁK, D. Polynomial chaos expansion for surrogate modelling: Theory and software. Beton und Stahlbeton, 2018, vol. 2, no. 113, p. 27-32. ISSN: 0005-9900.
Detail

Responsibility: Novák Drahomír, prof. Ing., DrSc.