Search

This paper proposes a methodology for using text analytics on computerized maintenance management system (CMMS) databases to benchmark the operational performance of commercial buildings. To this end, we extracted five years’ worth of service request and work-order logs from five large commercial buildings in Ottawa, Canada. We employed the association rule mining method on these datasets to identify building, system, and component-level recurring work-order taxonomies and common failure modes. The potential of Sankey diagrams, survival curves, and stacked line plots to effectively visualize the temporal, spatial, and categorical anomalies in the service request patterns was examined. It was identified that often only a few floors and service request types account for most of the service requests in a building. By applying the association rule mining algorithm on the work-order logs, it was identified that the lighting-related complaints were resolved by replacing ballasts and lights, and the thermal complaints were addressed by adjusting the temperature setpoints, airflow rates, and fan operation schedules. [ABSTRACT FROM AUTHOR]

The building energy management systems (BEMS) of many commercial buildings rely on constant thermostat set-points in their HVAC controllers, which often results in increased energy use and compromised occupant comfort. Therefore, occupant-centric controls (OCC) have gained increased research interest in reducing the energy consumption of buildings and improving occupant comfort. One of the goals of OCC is the development of smart thermostats that could learn from occupants' set-point adjustments to derive adaptive set-points. This research utilizes data from indoor environments and occupants' thermostat interactions to learn occupant temperature set-point preferences using a reinforcement learning (RL) algorithm. RL is a promising technique in this field because of its capabilities in learning user behavior without the need for occupant behavior and occupancy models. A tabular Q-learning algorithm is implemented using the Python API within an EnergyPlus simulation environment with an occupant behavior model. The objective is to learn the occupant behavior while minimizing energy use. Results demonstrated the ability of the developed RL algorithms to learn preferred set-points and adjust them accordingly, hence reducing occupant interactions. We propose future research for the implementation and field testing of smart thermostats. Such research would aid in shifting the process of maintaining occupant comfort from being the responsibility of building operators to being an automated process as well as achieving energy savings by taking decisions that consider long-term benefit. [ABSTRACT FROM AUTHOR]

Fault detection, diagnostics, and prognostics (FDD&P) can greatly help improve the performance of building operations by reducing energy consumption for heating, ventilation and air-conditioning (HVAC) while maintaining occupant comfort at the same time. In particular, prognostics as an emerging technique, is attracting an amount of attention from building operators and researchers because it enables a pro-active fault prevention strategy through continuously monitoring the health of building energy systems. In this paper, we propose to develop a machine learning-based method for HVAC prognostics. Building on techniques from machine learning and data mining, the proposed methods can help develop predictive models from the historic building operation and maintenance data. After presenting the proposed method, we discuss the building operation simulation conducted to generate data for evaluating the feasibility and usefulness of the proposed methods. The results from these numerical experiments demonstrated that the machine learning-based methods can be effective for HVAC prognostics. [ABSTRACT FROM AUTHOR]