Research

Developing computational framework combining optical satellite imaging and LiDAR for gas pipeline monitoring.

Publication: in preparation

Developing cyber twins using data from robotic systems (i.e., drones) with pure-vision 3D reconstruction (3D-Gaussian Splatting).

Publication: in preparation

Proposed acoustic sesing for the impurity gas sensing for spent nuclear fuel (SNF) canisters. Applied MLs algorithms to solve the forward and inverse problems in impurity gas sensing.

Publications: [review] [paper 1] [paper 2] [paper 3] [paper 4]

Figure source: [paper 1]

Developing RAG using open-source LLMs for analyzing cost factors in large public transit projects. The purpose is to understand why transit project is expensive in USA.

Publication: in preparation

Developed generative machine learning (diffusion models, conditional variational autoencoders) for design of structural cross-sections.

Publications: [paper 1] [paper 2]

Figure source: [paper 1]

Developed ML pipelines with time-series decomposition to predict the decay heat of used nuclear fuels in actual dry cask up to year 2050.

Publications: [paper 1]

Conducted dynamic modal testing on SNF canister mock-up and developed ML to identify and localize fuel assemblies damage using the frequency response function.

Publications: [paper 1] [paper 2] [paper 3]

Figure source: [paper 1]

Developed Rubber-sleeved Stud connectors and conducted push-out tests and non-linear finite element analysis using ABAQUS.

Publications: [review] [paper 1] [paper 2] [paper 3]

Figure source: [paper 1]

Quantified self-healing performance of RC beams with PU-filled engineered aggregates.

Publications: [paper 1]

Developed explainable ML to assess the conditions of 1,000+ telecommunications tower.

Publications: [paper 1]