top of page
20230226_090342.jpg

NIKHIL MAHAR 

PhD Scholar 

Nikhil Mahar is from Uttarakhand and has completed his bachelors from Graphic Era Hill University, Dehradun. His area of research is structural health monitoring of structures utilizing deep learning.

  • LinkedIn

Area of research

Structural health monitoring of structures utilizing deep learning

EMAIL

d22123@students.iitmandi.ac.in

EDUCATION

2022 - Present                                        


2018-2020                                               

2013-2017                                           
 

IIT Mandi, IN

NIT Meghalaya, IN

Graphic Era Hill
University Jaypee IN

PhD in Structural and Earthquake Engineering
M.Tech in Structural Engineering
B-Tech in Civil Engineering

PROJECTS 

Picsart_23-05-01_15-53-58-353_edited.jpg

1

Master’s Thesis:
Damage detection in RCC beam with ANNs

Utilized artificial intelligence to recognise crack pattern on RCC beam model under load. Predicted crack
patterns were compared with orignal crack patterns developed in finite element analysis software.

2

Bachelor’s Thesis:

Earthquake resistant design of multi-storey building

Systematic designing of all members of multi-storey building was done as per IS codes.

3

Approximate solution of Burger’s equation using Physics informed neural network

Solution of Burger’s equation without utilizing data in neural networks.

4

Application of hybrid PINN towards solving ODEs.

Hybrid PINN approach in an RNN cell a data-driven stress intensity range model is combined with a
physics based crack length increment model to study the fatigue crack growth with far-field cyclic stress
time history.

5

RK-Scheme to solve ODE using PINN

Both forward and inverse problems are solved by assisting Runge-Kutta method within neural network.

6

Reconstruction of signal

Reconstruction of signal from sparse data of single degree of freedom linear and non-linear systems using
physics informed neural networks (PINN)

Picsart_23-05-01_15-53-58-353_edited.jpg

7

BVP solution with PINN

Solution of a cantilever beam under point load at end using physics-informed neural networks(PINN)

8

Custom Recurrent Neural Network

Building many to many custom RNN for predicting time domain data.

9

Finite difference layer embedded neural network

Replacing adagrad with finite difference layer to solve 1st order linear ODE with physics informed autore-
gressive model.

10

Multi-fidelity transfer learning for consolidation of soil

A project undertaken to determine consolidation of soil by developing multi-fidelity PINN transfer learning
framework.

11

Girsanov based EM scheme for non-linear SDEs

Determined trajectories of non linear stochastic SDOF and MDOF systems by using Grisanov based Euler
Maruyama scheme.

PUBLICATIONS

2021

Damage detection in RC beam utilizing feed-forward
backpropagation neural network technique

In: Asian Journal of Civil Engineering 22.8, pp. 1551–1561

Mahar and Debabrata Podder

2020

Damage detection in RCC beam with ANNs
Thesis for M.Tech

Mahar and Debabrata Podder

INTERESTS

  " You will never know what you're good at until you try"

Picture2.png

MORE TO EXPLORE

  • LinkedIn
  • Facebook

i4S

information and Statistics for
Stochastic Structural Systems

Location : i4s Laboratory (First Floor), A-14 Building, School of Civil & Environment Engineering, North Campus, IIT Mandi, Himachal Pradesh 175075, India

 

Email : teami4s.iitmandi@gmail.com

Phone01905 267 261

bottom of page