A fuzzy sensor for measuring the mixed air temperature in air-handling units

The paper describes the development of a fuzzy sensor for measuring the average temperature of the air leaving the mixing-box of an air-handling unit. This measurement is often required by schemes for automatically detecting the presence of faults in air-conditioning systems. The measurement of the mixed air temperature is notoriously difficult because of spatial variations in the air temperature and velocity immediately downstream of the mixing-box. Variations in the sensor bias are investigated using a computer simulation of a typical mixing-box based on computational fluid dynamics. The simulation data are used to train a fuzzy relational model that relates the measured temperature to the average temperature of the air. The resulting fuzzy sensor takes bias errors into account and provides an indication of the uncertainties associated with the estimated average temperature at the current operating conditions. Results are presented that show how auxiliary information can be used to reduce the measurement uncertainties.