Scalable Cloud-Based Load Management System, Learning Models, and Optimization Framework for Connected Communities

As we are going through the electrification of transportation and building sector, the pattern of the load is changing drastically. Meanwhile, the penetration of renewable generation has changed power generation from centralized plants to distributed energy resources. The electric grid needs to adapt to these changes to ensure reliable operation. This requires integrating and coordinating resources at scale. Buildings are a major energy consumer in the United States and are responsible for 41% of electrical energy. Understanding buildings’ energy consumption patterns and enabling them as a grid asset is crucial for providing grid services while ensuring building occupant comfort. To achieve this, there is a need for an energy management system that can collect and manage data from a pool of heterogeneous devices in buildings and optimize their energy usage. In this dissertation work, a comprehensive literature review of available home energy management system (HEMS) is provided, and the challenges and opportunities associated to HEMS are identified. Based on this literature review and identified gaps in the state-of-the-art HEMS, a cloud-based and reliable HEMS is designed and implemented to optimize the energy use of devices in residential buildings. This HEMS is field validated at a connected neighborhood. A load forecasting algorithm is designed, developed, and integrated into this HEMS to model and forecast the system load variations. This forecast is fed to optimization to improve its accuracy and performance. Furthermore, a novel linear multi-objective model predictive control algorithm is introduced to create the optimal vendor-defined operational mode schedule for integrated Photovoltaic and energy storage system (ESS). Integrating operational modes into optimization will lead to a non-linear optimization model that cannot be easily solved using open-source solvers and is impractical for real-world deployments due to high computational requirements. The proposed novel approach incorporates the ESS operational modes into the optimization and links them to overall system charge and discharge power while keeping the optimization linear and easy to solve at the edge. This dissertation studies the design and implementation of a cloud-based HEMS along with forecasting algorithms and optimization developed and integrated in this HEMS to improve energy efficiency and reduce energy costs.