| Summary: | The energy consumption of HVAC has exceeded over 50% in the building sector over the past years [2]. Among all components of the HVAC system, the water chiller is the most energy-consuming component. Any fault or failure in the chiller may lead to further increase in energy consumption and inefficiency, higher O&M cost, and release more heat into the environment, thus contributing to global warming. Consequently, fault detection and diagnosis (FDD) had been proposed to detect and diagnose the possible fault types and their respective root causes. Furthermore, the implementation of FDD helps to identify the faults automatically and in a timely manner so that the user can resolve the fault as soon as possible for their HVAC system. Early detection and diagnosis using these methods prevent consequential harm and other mechanical damages on the chiller. The main focus of this project will be on using a Long Short-Term Memory (LSTM) to develop an FDD model. LSTM will target directing and diagnosing seven typical faults found in HVAC systems.
|
|---|