Browsing by Author "Dewapriya, MAN"

Now showing 1 - 3 of 3

item: Article-Full-text
Characterizing fracture stress of defective graphene samples using shallow and deep artificial neural networks
(Elsevier, 2020) Dewapriya, MAN; Rajapakse, RKND; Dias, WPS
Advanced machine learning methods could be useful to obtain novel insights into some challenging nanomechanical problems. In this work, we employed artificial neural networks to predict the fracture stress of defective graphene samples. First, shallow neural networks were used to predict the fracture stress, which depends on the temperature, vacancy concentration, strain rate, and loading direction. A part of the data required to model the shallow networks was obtained by developing an analytical solution based on the Bailey durability criterion and the Arrhenius equation. Molecular dynamics (MD) simulations were also used to obtain some data. Sensitivity analysis was performed to explore the features learnt by the neural network, and their behaviour under extrapolation was also investigated. Subsequently, deep convolutional neural networks (CNNs) were developed to predict the fracture stress of graphene samples containing random distributions of vacancy defects. Data required to model CNNs was obtained from MD simulations. Our results reveal that the neural networks have a strong ability to predict the fracture stress of defective graphene under various processing conditions. In addition, this work highlights some advantages as well as limitations and challenges in using neural networks to solve complex problems in the domain of computational materials design.
item: Article-Abstract
Effects of Large Retarder Overdose on Concrete Strength Development
Dias, WPS; Dewapriya, MAN; Edirisooriya, EACK; Jayathunga, CG
Retarders are used to delay the setting time of concrete. Retarder overdosing is not an uncommon problem, but its effects are not well understood. In this study, laboratory tests were conducted to determine the effect of retarder overdosing on the setting time and strength of concrete, using both cubes, and cores from a large specimen. It was found that overdose levels of 3 times the normal dosage had little effect on setting and strength development. Also, concrete with even 6 times the normal dosage of retarder did eventually set and gain strength. Tests were also conducted to determine cement setting time under different temperature and surface drying (wind) conditions, in order to make inferences regarding strength development of concrete surfaces compared to interiors.
item: Thesis-Abstract
Exploring patterns in historic earthquake and Tsunami data
Dewapriya, MAN; Dias, WPS
Gusiakov has compiled a comprehensive database compiled that gives data such as earthquake intensity, geographical location, resulting tsunami intensity etc. There have been various relationships proposed between earthquake and tsunami intensities, including by Gusiakov, who also proposes that regional effects may affect this relationship. Meanwhile, artificial neural networks (ANNs) have become a very powerful way of establishing input-output relationships, though lacking the formality of multiple regression (MR) techniques. Various approaches have been used to minimize the "black box" nature of ANNs, including the use of sensitivity analysis. Adaptive network based fuzzy inference systems (ANFIS) are a newly emerging alternative to ANNs. In this work, Artificial Intelligence (AI) methods (ANN and ANFIS) along with MR analysis were used as tools to explore the patterns in historic earthquake and tsunami data. The accuracy of the three modeling schemes were compared and sensitivity analyses performed. Vulnerability curves have been developed using Monte Carlo simulation that reasonably match survey based curves for the vulnerability of coastal houses to tsunami wave height. This Monte Carlo simulation was used in this work to establish the resulting reduction of vulnerability if proposed strengthening techniques are adopted.