Research

Timeline

3-year smart home project with community partners (IHCDA, BWI, Purdue)
A major student for idea development, figure works, and proposal writing assistance.
Project management role such as sensor procurement, meeting arrangement, communication between multiple partners/students/faculties.

Prototyping and choosing WiFi-based thermostats and circuit-level power sensors and their data collection to a database.
Installation and trouble shooting of sensors for 50 residential units located in Indianapolis, IN.
Documentation of data communication and database schema to work with software developers to accommodate multiple sensors communication in real-time.
Development of R (Shiny) and HTML based interactive data analytics tool for time-series data (for internal-use).
Procurement/installation/design of Tablet/Smart speaker for smart home interface (SmartE) with research team.

“A data-driven building energy normalization model for eco-feedback design in smart and connected multi-family residential buildings.”
Energy comparison of residential units in a multi-family residential building as an eco-feedback.
Normalized groups by considering inter-unit heat transfer and unobserved disturbances.
Fair comparison of the impact of thermostat behavior on heating/cooling energy consumption.
Online model update via Bayesian inference to be used for eco-feedback application.
Conference paper: PURDUE 2018.

“Online building energy model to evaluate heating and cooling-related behavior changes for eco-feedback in a multifamily residential building.”
Unit-level heating and cooling energy prediction model for actionable eco-feedback design a multi-family residential building.
A single zone gray-box model from thermostat and heat pump power measurement only.
Online parameter learning with state filtering to be used without a whole year dataset in Bayesian framework.
Conference paper: IBPSA 2019.

“A Scalable Building Modeling Approach using Probabilistic Deep learning (tentative).”
A household-level heating and cooling energy prediction model using Deep neural network.
Overcoming the limitation of black-box model through Bayesian update with pre-trained model to be used any types of residential unit.
Generation of multiple simulated data (EnergyPlus) with stochastic noise (Python EnergyPlus plugin) through Eppy package.
Solving probabilistic deep learning model by using Pyro and Pytorch.
In preparation.

Annual energy simulation for liquid-desiccant and dew point evaporative cooling based 100% outdoor system.
Implementation of PI controller, empirical liquid desiccant and dew point evaporative cooling model in a BCVTB co-simulation framework with EnergyPlus to test a new control algorithm (variable air volume and temperature control).
Development of a data-driven empirical model of PEM fuel cell system to be used for LDEOS system as a cogeneration application.
Conference paper: ASHRAE 2016.
Journal papers: ENB 2016 ENB 2015.

Design of environmental chamber to test the performance of indirect evaporative cooler.
Installation of sensors and data collection platform on PLC-based SCADA machine including wet-bulb temperature, airflow chamber.
Design/installation of indirect evaporative cooler with a manufacturing company.
A numerical model of indirect evaporative cooler by using finite difference method to design the prototype.
Journal papers: ENERGY 2016, APEN 2017

Development various air-side economizers and their annual energy simulation model for data center cooling application.
Development of a simplified server model that can be used for data center cooling energy simulation.
Estimating the impact of air containment in a modular data center through one-to-one scale test bed experiment and CFD simulation.
Development of excel VBA-based tool to estimate rough annual energy consumption of data center cooling for practical use.
Journal papers: APEN 2015, ATE 2015, ENB 2015, ATE 2016.