Data, Energy, Privacy Management Techniques for Sustainable Microgrids

Abstract

The increased reliance on burning fossil fuels to generate electricity is rapidly depleting our planets finite resources and contributing to climate change. Recent studies show that buildings consume 40% of the annual energy consumption. Consequently, techniques to make buildings self-sustainable, while ensuring user well-being and comfort, are crucial for achieving a sustainable energy future in smart cities. Recently, the growing adoption of renewable energy sources has shifted the emphasis from large-scale centralized utility control of power grids to more localized energy system in microgrids, which comprises of residential and commercial buildings; and generate, store, and share electricity to balance local generation and consumption. In this dissertation, we investigate the feasibility of microgrids by addressing three main challenges: i) how to collect and manage energy data from microgrids; ii) how to conduct energy energy management based on data analytics to minimize the total cost; iii) how to protect users? privacy in microgrids. Specifically, we first designed a low-cost hardware PowerQM to enable accurate power quality monitoring. Next, to collect energy data from deployed multiple energy meters in residential homes, we present E-Sketch, a middleware for utility companies to gather data from smart meters with much less storage and communication overhead. Then, based on the energy data collected, we propose M-pred for accurate demand forecast by utilizing high granularity data collected from residential homes. With M-Pred, we propose two energy management techniques in microgrids: i) exploring energy sharing among residential homes to improve the utilization efficiency of renewable energy (e.g., solar energy); ii) scheduling demand in residential homes and generation in microgrids to minimize the total operational cost. Finally, to protect data privacy of homeowners, we leverage the unique feature of hybrid AC-DC microgrids and propose Shepherd, a privacy protection framework to effectively protect occupants privacy. We implement and evaluate our proposed management techniques based on large energy datasets collected from more than 700 residential homes. The evaluation results show that i) our power quality meter PowerQM can achieve similar or even better accuracy than existing commercial products with much lower cost; ii) E-Sketch can significantly reduce the required data storage by 90% while preserving the data accuracy with more than 99%; iii) M-pred can achieve accurate demand forecast with negligible errors (e.g., Mean Absolute Percentage Error is 2.12%); iv) energy sharing can improve utilization efficiency of renewable energy up to 30%, while demand and generation scheduling can reduce total operational cost by 30% in microgrids; v) Shepherd can effectively protect users privacy of energy consumption information from multiple detection algorithms.