THE ONLY WAY TO DO GREAT WORK IS TO LOVE WHAT YOU DO
THE ONLY WAY TO DO GREAT WORK IS TO LOVE WHAT YOU DO
1) Undergraduate Thesis: National Institute of Technology, Warangal
Guide: Dr. Naresh Thota
1) Built an LSTM Recurrent neural network on the Tennessee Eastman Process data set and obtained
the confusion matrix and calculated the Fault detection rate.
2) In addition, built a Machine learning model using Random Forest and XgBoost and obtained an accuracy of 89.86% and 92.88%.
Achievement: Among the top 3 students out of 110 in the Department of Chemical Engineering, NIT Warangal to receive the Highest Academic Grade for demonstrating exemplary work in the Final year project.
2) Effect of weak inertia on the chaotic dynamics of periodically forced neutrally buoyant spheroids in simple shear flow. - IAS Fellowship
Guide: Dr.T.R Ramamohan
Worked with Dr. T R Ramamohan ( Emeritus Professor) on the topic Effect of weak inertia on the chaotic dynamics of periodically forced neutrally buoyant spheroids in simple shear flow
Formulated the non-linear coupled higher-order differential equation due to the addition of external torque on prolate spheroids and solved them using MATLAB and R and obtained the phase space plots (attractors) to analyze the chaotic nature.
Furthermore, we calculated the Lyapunov exponents and correlation dimension to determine the chaotic nature. In addition to it, we analyzed the effect of chaos and sensitivity of aspect ratio on small Reynolds Number. Finally, we concluded that increasing the small Reynolds number will make the system stable and obtained the critical Reynolds to know the transition of chaotic behavior to the stable solution of the system.
The Manuscript was submitted and accepted for International Conference on Applied Maths and Non-linear Dynamics.
3) Data Analysis of Tennessee Eastman process
Guide: Dr.Nabil
Institute: Indian Institute of Technology, Tirupati, India
Worked remotely with Dr.Nabil on Data analysis of the Tennessee Eastman process. In this project, I worked on data interpretation, data pre-processing, and data visualization of the Tennessee Eastman process dataset where I plotted time series plots of inputs and outputs of this process to do further analysis.
In this project, I built a model based on Machine learning algorithms, especially focusing on Kernel-based Gaussian regression, Linear Regression, in MATLAB to observe the variation of process variables with respect to output( Product G) and (Product H). Based on the models built for product G, I got an accuracy of 49% with Gaussian process regression with the exponential kernel being used and for product H, I got an accuracy of 45% with multiple linear regression.
Some of the results that we concluded from the model are shown in the form of plots including regression plots and time series analysis of input, output, and flow rate of components in the process.
4) Linear and Non-linear stability analysis of convective flow in a porous media
Guide: Dr. P.A Lakshmi Narayana
Institute: Indian Institute of Technology, Hyderabad, India
Worked remotely on Linear and Non-Linear stability analysis of Convective Flow in a porous media(Hydrodynamic stability).
In this project, I modeled and solved Higher-order Coupled Differential equations that came from the literature on Convective Flow in a porous media and applied the concept of perturbation parameters and solved those equations in MATLAB using boundary conditions and got the results for Marginal stability analysis which depends on Rayleigh number and Peclet number.
You can mail me to know more about how to solve ordinary differential equations with boundary conditions using MATLAB and to know more about the work.
5) Design Optimization of Shell and Tube Heat Exchanger
Guide: Dr.Naresh Thota
Institute: National Institute of Technology, Warangal, India
In this project, we implemented a Multi-objective optimization of shell and tube heat exchangers based on cost and length and solved using a multi-objective genetic algorithm using MATLAB. Designed a heat exchanger based on the design variables baffle spacing, outer and shell diameter. We got the final optimized cost is 53,248/- with a length of 3.782m. We concluded from the project that both the cost and length of the heat exchanger are opposite entities means an increase in one always leads to a decrease in other functions. Hence we can say that our three design variables are non-inferior solutions.