A neural network regression model for estimating the lifespan of a Fibre Bundle

Abstract

Abstract Fibre Bundle Models (FBMs) use generalized distributions like the Weibull distribution to study the failure mechanics of disordered material under different load-sharing schemes. Here we attempt to use a simple neural network regression model to estimate the lifespan of Fibre Bundles for axial loading under the Global Load Sharing (GLS) scheme. We find that using neural networks can give a reliable estimate (within ∼2%) of the lifespan for different initial conditions. We also develop a semi-analytical expression for the lifespan of a bundle of fibres. The aim is to establish an empirical relationship using a neural network regression (NNR) method that helps us estimate the ultimate tensile strength. The expressions and methods developed here can be a precursor to future investigation under those cited in the following section(s).