Graduation Year

2023

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Electrical Engineering

Major Professor

Yasin Yilmaz, Ph.D.

Committee Member

Ismail Uysal, Ph.D.

Committee Member

Mahshid Rahnamay Naeini, Ph.D.

Committee Member

Ankit Shah, Ph.D.

Committee Member

Lu Lu, Ph.D.

Keywords

EV charging, home energy management system, hospital capacity, Resource allocation, sea level rise

Abstract

Optimization, which refers to making the best or most out of a system, is critical for an organization's strategic planning. Optimization theories and techniques aim to find the optimal solution that maximizes/minimizes the values of an objective function within a set of constraints. Deep Reinforcement Learning (DRL) is a popular Machine Learning technique for optimization and resource allocation tasks. Unlike the supervised ML that trains on labeled data, DRL techniques require a simulated environment to capture the stochasticity of real-world complex systems. This uncertainty in future transitions makes the planning authorities doubt real-world implementation success. Furthermore, the DRL methods have limitations for different application environments; slow convergence, unstable learning, and being stuck in local optima are a few of them.

We address these challenges in our environmental, healthcare, and energy systems projects by carefully (1) modeling the system dynamics we achieved through research and collaboration with domain experts and (2) state-of-the-art DRL techniques for experimental analysis. Our experimental results and comparative analysis with the other optimization methods demonstrate the efficacy of DRL-based techniques. The success lies in appropriately modeling the critical decision-making features, reward function, and state transitions. In the process, we have developed novel DRL (Multi-agent and Multi-objective) algorithms.

Download

Share

COinS