Projects
Ongoing Projects
Completed Projects
A co-kurtosis PCA based dimensionality reduction with ANN reconstruction, IISc (2023)
Proposed the CoK-PCA-ANN dimensionality reduction method, which combines a CoK-PCA-based projection (encoding) and an ANN-based reconstruction (decoding) to capture stiff chemical dynamics in combustion datasets accurately.
Investigated the efficacy of the CoK-PCA-ANN method in four combustion test cases spanning conventional single-stage to complex two-stage ignition kinetics, different combustion regimes, as well as from a simple homogeneous reactor to a spatiotemporally evolving 2D flow.
Initial results have been accepted as abstracts for oral presentations at SIAM CSE'23 and APS DFD'23. Detailed results have been published as a journal manuscript in Elsevier Combustion and Flame.
Extreme event detection in hydrogen combustion relevant to reheat burners, IISc (2023)
Employed a co-kurtosis tensor-based anomaly detection algorithm to identify extreme events such as flame instabilities in reheat burners of hydrogen-fired sequential gas turbine engines.
Initial results have been accepted as a short paper for oral presentation at ASPACC'23. A related work in collaboration with TU Delft has been selected for presentation at the Combustion Institute's 40th International Symposium.
Use of Hidden Markov Models for keystroke biometric studies, IISc (2022)
In this project, we explored the use of a Partially Observable Hidden Markov Model (POHMM) which is an extension of Hidden Markov Model (HMM) where both hidden states and emissions depend on an observable independent Markov chain and use it for the classification of a user as an imposter or genuine user based on his/her keystroke dynamics. We have implemented POHMM and compared the results with HMM on the CMU keystroke benchmark dataset.
Numerical Solution of Solid-Solid Phase Change Material (SS-PCM) based heat sink, NITT (2019)
This project involved carrying out a numerical heat transfer study on a heat sink embedded with layered perovskite (SS-PCM) for finding an optimized fin configuration based on constant fin volume for circular, square, triangular and tapered triangular geometries.