Radial Basis Function Neural Network-Based Modeling of the Dynamic Thermo-Mechanical Response and Damping Behavior of Thermoplastic Elastomer Systems

Abstract

The presented work deals with the creation of a new radial basis function artificial neural network-based model of dynamic thermo-mechanical response and damping behavior of thermoplastic elastomers in the whole temperature interval of their entire lifetime and a wide frequency range of dynamic mechanical loading. The created model is based on experimental results of dynamic mechanical analysis of the widely used thermoplastic polyurethane, which is one of the typical representatives of thermoplastic elastomers. Verification and testing of the well-trained radial basis function neural network for temperature and frequency dependence of dynamic storage modulus, loss modulus, as well as loss tangent prediction showed excellent correspondence between experimental and modeled data, including all relaxation events observed in the polymeric material under study throughout the monitored temperature and frequency interval. The radial basis function artificial neural network has been confirmed to be an exceptionally high-performance artificial intelligence tool of soft computing for the effective predicting of short-term viscoelastic behavior of thermoplastic elastomer systems based on experimental results of dynamic mechanical analysis.

Modelling the Stiffness-Temperature Dependence of Resin-Rubber Blends Cured by High-Energy Electron Beam Radiation Using Global Search Genetic Algorithm

“…We proved in our earlier works [40][41][42] that the temperature dependence of the storage modulus E (T) at a given constant oscillating frequency of dynamic mechanical loading for thermoplastic elastomer systems within the whole temperature range of its service life can be quantitatively described with a high level of reliability by the Mahieux and Reifsnider's unified analytical model [43] based on a Weibull distribution of the failures of secondary bonds between macromolecular chains throughout the primary as well as secondary relaxation processes. The presented study focuses on the ability of such approach to predict the temperature dependence of dynamic mechanical behaviour of much more complex PMX3 resin-rubber polymeric system for the first time, cured by different doses of high-energy EB radiation, which represents, in terms of molecular forces, a specific blend of three different polymers of two different types-rather incompatible and immiscible hybrid blend of one elastomer with two relatively very high compatible and miscible thermosets-while its viscoelastic properties are determined not only by the temperature but also by the size of the applied radiation dose.…”

Artificial Immune System in Doing 2-Satisfiability Based Reverse Analysis Method via a Radial Basis Function Neural Network

“…We set w ji = 1 because other values of w ji will result in the biased selection, which leads to weighted 2SAT [4,13,61]. c i is the center and σ i is the width as shown in the following equations:…”

An advanced meta-learner based on artificial electric field algorithm optimized stacking ensemble techniques for enhancing prediction accuracy of soil shear strength

“…It is noted that the model construction phase of the RBFNN model requires a proper determination of the number of neurons (N n ) and the RBF width (σ). An appropriate set of N n and σ ensures the success of establishing a [51,52]. Additionally, the parameter σ affects the influence of RBF nodes on each data point; therefore, it determines the generalization of the prediction model.…”