A fuzzy approach to fault diagnosis in the presence of sensor bias

Describes a fuzzy approach for rule-based fault diagnosis in the presence of sensor bias. The causes of sensor bias and its effect on fault diagnosis schemes are presented. The consequences of assuming that the bias is uniform and varying are compared and contrasted. A fuzzy model-based fault diagnosis scheme is modified to take account of varying sensor bias. The proposed scheme is applied to an air-conditioning system. Results are presented that show that better diagnostic results are obtained than when a constant worst-case value of the bias is assumed