Fault detection and diagnosis on HVAC variable air volume system using artificial neural networks

Abrupt  faults  on  HVAC  components  as  blocked dampers or broken fan belt can be successfully detected by methods based on logic rules. On the other hand, those method are less efficient to detect fouling on coil or scaling in tubes that are progressively decreasing the energy efficiency and are long-lasting phenomena. Previous work on simulation data shows that methods based on artificial neural networks (ANN) are adapted to solve this problem. Model method consists in comparing real behavior of the HVAC plant to a normal behavior given by ANN trained during a preliminary phase. The main difficulty of using ANN for fault detection is to produce the training data. Indeed, the performance of the detector is linked to the quality of these data. The procedure of using real data obtained after a recommissioning is really problematic. An alternative way using a physical model is tested to produce training data for the cooling coil. This model of cooling  coil requires only a rating point to be characterized. ANN performance with training on simulation data is evaluated on a VAV system. Artificial faults are introduced in the real plant to simulate standard faults occurring in building HVAC system.