Event Title

Development of Deep Learning based model to predict Dynamic Modulus in Matlab using Artificial Neural Networks (ANN)

Presenter Information

Mohammed Moinuddin

Loading...

Media is loading
 

Faculty Mentor

Dr. Mena Souliman

Document Type

Poster Presentation

Date of Publication

January 2021

Abstract

Dynamic modulus (|E*|) is a crucial engineering property used to determine the performance and stiffness characteristics of asphalt pavements. Various models have been developed to predict the Dynamic modulus like Original Witczak equation, Modified Witczak Equation, Hirsch model, Law of mixtures parallel model, ANN model and Resilient modulus-based model. All these models are predictive models used to estimate the Dynamic modulus as it is a cumbersome and considerably time-consuming process to measure the Dynamic modulus in lab due to various issues like preparation of samples, availability of expensive equipment, and skilled personnel to perform tests and to equilibrate temperature. In this study, we utilize the data which was used to develop Modified Witczak equation to develop an equation in Matlab using ANN. All the factors affecting the Dynamic modulus has been identified and regression analysis has been performed on the data points to generate best equation. In total, 7400 data points was used from 346 mix samples to generate the equation. The main aim of the study is to make an effort to overcome the shortcomings in the previous predictive models, which are known for being inaccurate and accuracy have been questioned by many researchers. Efforts are to develop a simple mathematical equation to determine Dynamic modulus, which requires to perform trial and error technique for development of optimized network structure.

Keywords

Dynamic modulus of asphalt concrete, Matlab, ANN.

Persistent Identifier

http://hdl.handle.net/10950/3067

Dynamic Modulus Lyceum Poster.pdf (856 kB)
Dynamic Modulus Poster PDF

This document is currently not available here.

Share

COinS
 

Development of Deep Learning based model to predict Dynamic Modulus in Matlab using Artificial Neural Networks (ANN)

Dynamic modulus (|E*|) is a crucial engineering property used to determine the performance and stiffness characteristics of asphalt pavements. Various models have been developed to predict the Dynamic modulus like Original Witczak equation, Modified Witczak Equation, Hirsch model, Law of mixtures parallel model, ANN model and Resilient modulus-based model. All these models are predictive models used to estimate the Dynamic modulus as it is a cumbersome and considerably time-consuming process to measure the Dynamic modulus in lab due to various issues like preparation of samples, availability of expensive equipment, and skilled personnel to perform tests and to equilibrate temperature. In this study, we utilize the data which was used to develop Modified Witczak equation to develop an equation in Matlab using ANN. All the factors affecting the Dynamic modulus has been identified and regression analysis has been performed on the data points to generate best equation. In total, 7400 data points was used from 346 mix samples to generate the equation. The main aim of the study is to make an effort to overcome the shortcomings in the previous predictive models, which are known for being inaccurate and accuracy have been questioned by many researchers. Efforts are to develop a simple mathematical equation to determine Dynamic modulus, which requires to perform trial and error technique for development of optimized network structure.