Reliable process data are the basis for efficient process evaluation, operation and control. When flawed information is used for process monitoring and control, the state of the system is misrepresented and the performances of the control and fault detection schemes may be poor. Therefore, to improve the estimation accuracy and reduce measurement uncertainties is essential for the optimal operation of intelligent monitoring, control and fault detection schemes in all kinds of systems. The work described in this thesis is aimed at tackling a particular kind of measurement error: systematic error (or sensor bias) that cannot be removed by calibration or simply installing a more accurate sensor. This kind of systematic error can occur when a sensor is not directly measuring the quantity that needs to be measured, for example, when the measured quantity varies spatially or when the sensor is not positioned at an appropriate location. The method adopted in this thesis uses Computational Fluid Dynamics (CFD) simulation to estimate the systematic errors caused by spatial variations and data fusion to combine the direct measurements with other knowledge to reduce measurement uncertainties which include both random errors and systematic errors. The application of this method in airconditioning systems has been described. The control signals for the actuators of the fans and the mixing box dampers, which are available through the building energy management system, can be used to estimate the airflow rates. CFD simulations are carried out to examine the relationship between the values of the systematic errors on the measurements of the mixed-air flow rate and temperature and the operation conditions. Results obtained from the simulation of a symmetric mixing box are used to derive mathematical models that can be used to estimate the measurement biases. The uncertainties associated with the CFD simulations and the bias estimators are evaluated. A method of fuzzy data fusion, which incorporates a compatibility function, is proposed. Both a statistical data fusion scheme and a fuzzy data fusion scheme are used to reduce the estimation errors associated with the supply, return and inlet airflow rates and the mixed-air temperature in air-conditioning systems. The estimates obtained from the data fusion schemes are used in a fuzzy fault detection scheme for the cooling coil. Promising results are obtained. It is shown that the use of the data fusion schemes, together with the estimation of the systematic errors, can greatly reduce the overall measurement uncertainties and improve the performance of the fault diagnosis scheme when the estimates the systematic errors are unbiased. It is concluded that fuzzy data fusion should be used if the systematic errors are varying and the estimates of the systematic errors are themselves biased.