Non-intrusive fault detection of reverse cycle air conditioning systems – Dissertation

Connor Gregory

The influence from buildings on global energy consumption, both residential and commercial, steadily increases each year. Growth in population, demand for building comfort and climate control ensure buildings will continue as a prominent energy consumer far into the future. Equating to over half a buildings total energy consumption, heating ventilation and cooling systems (HVAC) often operate in substandard conditions, decreasing efficiency by 15% to 30%. Much of this efficiency could improve with the widespread adoption of automated condition-based maintenance. Automated fault detection and diagnostic systems (FDD) provide a cornerstone for condition based-monitoring within HVAC systems. However, the current available FDD approaches employ invasive installation techniques, requiring fixing of various mechanical sensors to the pre-existing unit. With new research in the field of system monitoring via electrical measurements, this paper will explore a less invasive approach to HAVC fault detection. Using only electrical measurements, air-conditioning systems will undergo testing with simulated fault conditions. Machine learning algorithms are studied to predict the fault status, ultimately providing the client with appropriate diagnostic information. Providing fault detection capabilities via electrical measurements ensure a much less intrusive approach and does not restrict the measurement equipment to be local to the system. Ultimately the system would become far cheaper and convenient than traditional systems promoting installation and increasing the HVAC system’s efficiency.

Non-intrusive fault detection of reverse cycle air conditioning systems – Dissertation

Files and links (1)

Abstract

Details

Metrics