Fault Detection and Diagnosis in HVAC Systems Using Diagnostic Multi-Query Graphs

Abstract

Modern heat, ventilation, and air conditioning (HVAC) systems comprise several electrical components prone to failures. These failures, when left undetected, lead to increased operational costs, excessive energy consumption, deterioration of equipment and general discomfort. For the detection of hazardous situations, e.g. a fire, defective components in the HVAC systems may overshadow the occurrence of the catastrophe or cause stressful situations with false alarms. In both cases, there is a possibility of damage to human life and property that could have easily been prevented with working detectors. In this paper, we propose a diagnostic query-based fault detection and diagnosis (FDD) technique to monitor faults in the HVAC systems in safety-critical and non safety-critical situations. We consider the critical situations (detection of fire) time-sensitive and our technique determines whether an actual disaster has occurred or the system itself has malfunctioned within the system defined time-bound. Our technique ensures that the HVAC system gives a time-critical and reliable response for detecting a disaster in the building. Fault detection in non-critical conditions has no time restriction and is performed to minimize the energy consumption of the HVAC systems. We have used Fault Diagnostic Queries (FDQs) that are realized on the features and symptoms extracted from the sensors in the HVAC system. Using these FDQs, we formulate a Diagnostic Multi-Query Graph (DMG) for the structured and timely analysis of faults. To ensure that we meet the stringent timing constraints of the system, we optimize the DMG using a Genetic Algorithm (GA). The optimized DMG serves as the input for the scheduler that gives the execution time for fault detection and generates an optimal plan for the execution of the DMG. We have tested our (FDD) technique on buildings with 20, 40, 60 and 80 rooms, respectively. The fault detection in safety-critical situations in HVAC system has not been considered before in literature.