Sensor fault detection in building energy management systems
(EN)
Πουλιεζος Αναστασιος
(EL)
Κολοκοτσα Διονυσια
(EL)
Σταυρακακης Γεωργιος
(EL)
Pouliezos Anastasios
(EN)
Stavrakakis Georgios
(EN)
Kolokotsa Dionysia
(EN)
Πολυτεχνείο Κρήτης
(EL)
Technical University of Crete
(EN)
In this specific paper a methodology for detecting sensor failures in building energy management systems is presented. The fault diagnosis decision criterion is the average absolute prediction error between the actual and the predicted values of the sensor. Τhe predicted value is calculated by a model based on normal operation data. Three experiments are presented with simulated biases in the temperature, illuminance and CO2 sensors. Although the concept is simple, the results for fault detection are quite satisfactory.
(EN)
