Font Size: a A A

Machine learning based building control strategy

Posted on:2007-03-16Degree:Ph.DType:Thesis
University:Carnegie Mellon UniversityCandidate:Singhvi, VipulFull Text:PDF
GTID:2442390005972991Subject:Engineering
Abstract/Summary:
Meeting the user preferences of comfort, safety and privacy are the major factors in the success of many civil-infrastructure projects. In the building operation domain, user comfort and efficient operation have always been two primary, yet often competing objectives in developing building control strategies. There is a tradeoff-of-benefit in choosing to meet these goals simultaneously. In this thesis, we cast the problem of choosing a control strategy in a decision theoretic framework, where the objective is to minimize the building operation cost and simultaneously meet user comfort level at an acceptable comfort level. We present a building control strategy, implemented as an intelligent lighting control, that integrates user preferences, the state of the immediate indoor and outdoor environment, meets the goals of reduction in energy and addresses the complex tradeoff-of-benefits involved. We present results from three real life case studies. We also address the problem of daylight harvesting, and propose a novel sensor scheduling problem for wireless sensor networks. We include results from a user study, evaluating the effectiveness of the controlled strategy. The proposed control strategy can be successfully implemented in real environments to install intelligent lighting systems.
Keywords/Search Tags:Control strategy, Building control, User, Comfort
Related items