Fuzzy neural networks model for building energy diagnosis

A comfortable indoor climate environment is necessary for modern buildings and therefore the Heating, Ventilation and Air-Conditioning (HVAC) systems are widely used. Faults or disturbances are normally unavoidable in the systems and they will lead to more energy consumption or degradation of comfort level of indoor climate. Energy consumption is useful to detect the
faults. Fuzzy Neural Networks (FNN) model is presented and discussed in this paper. By applying the model on the measured data with fault of an open window in the room, the FNN model is shown to be a good candidate for energy diagnosis. In order to make the fault detection quantitatively, a threshold of a characteristic parameter, derivative of the multiplication of mean value and variance, is applied and shows good results.