Semantic Models and Reasoning for Building System Operations: Focus on Knowledge-Based Control and Fault Detection for HVAC

Abstract

According to the U.S. Energy Information Administration (EIA), the Building Sector consumes nearly half (47.6%) of all energy produced in the United States. Seventy-five percent (74.9%) of the electricity produced in the United States is used just to operate buildings. At the same time, decision making for building operations still heavily rely on human knowledge and practical experience and may be far from optimal. In a step toward mitigating these deficiencies, this dissertation reports on a program of research to identify opportunities for using semantic models and reason- ing in building system operations. The work focuses on knowledge-based control and fault detection for heating, ventilation and air conditioning (HVAC) systems. Decision-making procedures for building system operations are complicated by the multiplicity of participating domains (e.g., architecture, equipment, sensors, occu- pants, weather, utilities) that need to be considered. The key opportunity of this approach is a means to utilize semantic models for knowledge representation, inte- gration of heterogeneous data sources, and executable processing of semantic graph models in response to external events. The results of this dissertation are con- densed into three case-study applications; (1) Semantic-assisted model predictive control (MPC) for detection of occupant thermal comfort, (2) Semantic-based util- ity description for MPC in a chiller plant operation, and (3) Knowledge-based fault detection and diagnostics for HVAC systems.