The improvement on constitutive modeling of Nb-Ti micro alloyed steel by using intelligent algorithms. (15th February 2017)

• Record Type: Journal Article
• Authors: Wu, Si-Wei; Zhou, Xiao-Guang; Cao, Guang-Ming; Liu, Zhen-Yu; Wang, Guo-Dong
• Is Part Of: Materials & design. Volume 116(2017)
• Page Start: 676
• Page End: 685
• Publication Date: 2017-02-15
• Subjects: Nb-Ti micro alloyed steel -- Hot deformation behavior -- Flow stress -- Arrhenius-type model -- Genetic algorithm -- Artificial neural network; Materials -- Periodicals; Engineering design -- Periodicals; Matériaux -- Périodiques; Conception technique -- Périodiques; Electronic journals; 620.11
• DOI: 10.1016/j.matdes.2016.12.058
• Languages: English
• ISSNs: 0264-1275

Abstract

Abstract: The deformation behavior of Nb-Ti micro alloyed steel was experimentally obtained in the deformation temperature range of 900–1050 °C and strain rate range of 0.1–10 s − 1 . Base on the stress-strain curves, four constitutive models were established by using the modified forms of Arrhenius-type model considering compensation of strain or modified by intelligent algorithms such as artificial neural network and genetic algorithm. A comparative study has been made on the accuracy and effectiveness of the four models to predict the flow stress of Nb-Ti micro alloyed steels. The result shows that models modified by intelligent algorithms acquire higher accuracy than Arrhenius-type model considering compensation of strain. The model whose material parameters ( α, n, Q and ln A ) are modeled by artificial neural network performs better than the model whose material parameters ( α, n, Q and ln A ) are optimized by genetic algorithm. Among all the models, Arrhenius-type model considering compensation of strain with errors revised by artificial neural network obtains the highest accuracy. Graphical abstract: Highlights: The hot deformation behavior of Nb-Ti micro alloyed steel was studied by hot compression tests. The Arrhenius-type model considering compensation of strain was established. Artificial neural network was applied to model material parameters in Arrhenius-type model. Genetic algorithm was used to optimize material parameters in Arrhenius-type model. The constitutive models optimized by intelligent algorithms were compared and analyzed.