UCLA

With the increase in electric vehicle (EV) adoption in recent years, the impact of EV charging activity to the power grid has become increasingly significant. Although an EV is considered beneficial to the environment by reducing greenhouse gases, large amounts of un-coordinated EV charging could be detrimental to the power grid and thereby degrade power quality. Recent developments in Vehicle to Grid (V2G) technology has converted an EV to a distributed energy resource (DER). A modern smart grid with intelligent IoT devices, solar generation and battery storage provides additional opportunities but also additional challenges to the grid operator. To alleviate the negative effects of massive EV charging load and turn them into grid assets, the current dissertation performs research in designing and developing optimal EV charging strategies to integrate EVs into the smart power grid.

Using the UCLA Smart Grid Energy Research Center (SMERC) smart EV charging network infrastructure as the testbed, data has been collected regarding EV driver charging behavior for five years. Based on historical charging records, both deterministic and generative EV user behavior models are proposed to combine statistical analysis and machine learning to predict day-ahead EV driver itinerary and energy demand. Optimal Vehicle Grid Integration strategy is designed to realize different objectives including EV charging cost minimization, power grid stabilization, computational burden decentralization, increasing convergence speed, mitigating solar over-generation, etc. A distributed optimal bi-directional charging scheduling algorithm with asynchronous converging feature has been designed for load curve flattening; A two-stage optimization and a distributed water-filling algorithm have been developed for aggregating EVs to participate in energy market and demand response program. Both large-scale simulation and real-world implementation are conducted to validate and evaluate the performance of these algorithms. Results show that the proposed distributed optimal bi-directional charging scheduling algorithm is able to flatten power peak load by 35% when implemented in a test-bed located within the parking structure 9 in UCLA. A daily energy cost saving of 18% is achieved when the two-stage optimization algorithm is performed to control the EVs in a parking structure in the Civic Center Garage of the City of Santa Monica to participate in wholesale energy markets. Smart meter data collected in the Santa Monica parking lot shows the proposed charging control algorithm is able to mitigate the solar over-generation in the building by 50% on a daily basis. It can be concluded that our Vehicle Grid Integration strategy is effective in stabilizing power grid load, reducing charging cost and solving solar power over-generation problem. In addition to the development of EV user behavior models and Vehicle Grid Integration strategy, this dissertation also solves practical engineering problems for a scalable, reliable and safe EV bi-directional smart charging infrastructure.