Model-based Fault Detection and Diagnosis of HVAC systems using Support Vector Machine method

Preventive maintenance plays a very important role in the modern Heating, Ventilation and Air Conditioning (HVAC) systems for guaranteeing the thermal comfort, energy saving and reliability. Its key is a cost-effective Fault Detection and Diagnosis (FDD) method. To achieve this goal, this paper proposes a new method by combining the model-based FDD method and the Support Vector Machine (SVM) method. A lumped-parameter model of a single zone HVAC system is developed first, and then the characteristics of three major faults, including the recirculation damper stuck, cooling coil fouling/block and supply fan speed decreasing, are investigated by computer simulation. It is found that the supply air temperature, mixed air temperature, outlet water temperature and control signal are sensitive to the faults and can be selected as the fault indicators. Based on the variations of the system states under the normal and faulty conditions of different degrees, the faults can be detected efficiently by using the residual analysis method. Furthermore, a multi-layer SVM classifier is developed, and the diagnosis results show that this classifier is effective with high accuracy. As a result, the presented Model-Based Fault Detection and Diagnosis (MBFDD) method can help to maintain the health of the HVAC systems, reduce energy consumption and maintenance cost.