DETERMINATION OF ELASTOPLASTIC PROPERTIES OF IN-SERVICE PIPELINE STEEL BASED ON BP NEURAL NETWORK AND SMALL PUNCH TEST

In order to obtain the elastoplastic mechanical properties of in-service pipeline materials without shutting down transportation, this paper presents a method of combining artificial intelligence back-propagation (BP) neural network, small punch test and finite element simulation to obtain the elastic-plastic mechanical properties of materials by determining the true stress-strain curve of materials. Elastoplastic mechanical properties of X80 pipeline steel was obtained by this novel method. First, 457 groups of hypothetical material constitutive relations with different elastoplastic mechanical properties are obtained by systematically changing the parameters K and n in the Hollomon formula. The load-displacement curves of 457 groups of different materials were obtained by using the two-dimensional axisymmetric finite element model of small punch test with Gurson-Tvergaard-Needleman (GTN) damage parameters verified by experiments. The two groups of data are substituted into BP neural network and the corresponding relationship between the load-displacement curve of small punch test and the true stress-strain curve of the conventional uniaxial tensile test is established by the method proposed. This relationship can be used to obtain the true stress-strain curve of in-service pipeline steel based on the results of small punch test, so as to determine its elastic plastic mechanical properties. The accuracy and wide applicability of the BP neural network model are verified by comparing the true stress-strain curve of X80 pipeline steel obtained by BP neural network with the results of uniaxial tensile test and the experimental data of materials in the literature.