EAGER: User-Centric Interdependent Urban Systems: Using Multi-Modal Transportation Data for Demand Prediction and Management in Buildings

点击翻译按钮获取中文摘要

In 2015, 40 percent of total energy consumption in the U.S. was attributed to buildings. Building systems, such as heating, ventilation, lighting, steam and water supply systems, were the dominant end uses. Accurate predictions of utility demand are at the heart of efficiently managing and operating building systems for energy saving. This Smart and Connected Communities (S&CC) EArly-concept Grant for Exploratory Research (EAGER) project uses data in transportation systems to predict building occupancy and support real-time building system management. The rationale is that people use transportation and buildings sequentially, and the building occupancy (and to some extent utility demand) can be accurately predicted according to when and how many users are traveling and heading to buildings minutes or hours ahead. This theory will be tested in selected buildings on Carnegie Mellon University campus. The proposed research, if successful, creates a new paradigm for understanding the complex nature of interrelationships among various urban systems, using building systems and transportation systems as examples. It has great potentials to save energy and expand infrastructure life cycles, resulting in significant societal benefits. It will provide interdisciplinary training to students involved in this research on civil engineering, data science, facility management and control theory. The PIs will organize several web-based seminars to bring the theories and findings to a broader audience of students, researchers and urban system managers. The research results will be disseminated through journal publications, conferences, and workshops.The objective of this research is to fuse and analyze high-resolution real-time data regarding the usages of transportation and building systems and discover spatio-temporal correlations of user patterns among those systems. The spatio-temporal correlations would be critical, if discovered, for enabling cross-system demand prediction and management in an effective and efficient manner. In particular, this research gains a better understanding of the interdependency of roadway, transit, parking and building systems through holistically mining the massive data of all those systems. A critical component is to learn typical recurrent and non-recurrent flow patterns in a joint transportation-building network. In addition, this research develops a data-driven approach that automatically detects relations between building occupancy levels and required loads for various building systems to develop customized models for occupancy-load-based control. It also develops closed-loop control mechanisms for setback of building systems in real time.