Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Virtual In-Situ Calibration for Reliable and Resilient Sensing in Building Energy Systems
Advanced building energy systems for high performance and energy savings are required to use the information available from building sensor networks. Various data-driven technologies are also developed using the reliable training/testing measurement sets. If some sensors in the networks are erroneous, they will be ineffective. Conventional sensor calibration is conducted periodically to correct building sensor errors. Unfortunately, the conventional method can correct random errors, but cannot fix systematic errors due to the difference between working and calibration environments. Moreover, the conventional calibration has limitations in building energy systems because of the large-scale sensor network: (1) time and monetary cost; (2) disruption of normal system operation; (3) difficulty in accessing sensors embedded in equipment; and (4) the large number of sensors. Thus, it is important to develop a novel sensor calibration method that can handle the challenging systematic errors and the practical calibration issues in ensuring desirable building performance. This dissertation develops virtual in-situ calibration (VIC) to address that. Fundamental, applied, and experimental studies were conducted in order for VIC to be effective in real buildings. First, the VIC problem was mathematically formulated based on Bayesian inference to handle multiple sensors simultaneously by calibrating various random and systematic errors. Based on the formulation, several case studies were conducted to determine scientific issues causing the VIC failure and their mechanisms behind the negative impacts in real operational systems. Then, strategies for VIC were suggested to improve the VIC performance, overcoming the negative effects. Finally, the advanced VIC was applied to real building energy systems (unitary air conditioner systems). Considerable errors on the system energy balance, up to a 14.8% error, caused by sensor errors were identified and they were solved below 2% by the suggested VIC. These findings show VIC can have a high accuracy and stability in real buildings, providing a successful calibration in all working stages for an operational system. Consequently, the proposed VIC can be conducted with advanced building systems to provide reliable and resilient sensing environments for their sound operation. ^
Yoon, Sungmin, "Virtual In-Situ Calibration for Reliable and Resilient Sensing in Building Energy Systems" (2018). ETD collection for University of Nebraska - Lincoln. AAI10839617.