Sensor fusion and sensor fault detection with fuzzy clustering

In this paper, an efficient approach for multisensor fusion and fault detection is presented. This approach uses Fuzzy Cluster-Means (FCM) algorithm for signal separation. Subsequently, a fused signal is generated by the fusion engine based on the concept of the center of gravity (COG) deffuzification method. Sensor fault detector is designed based on the residual between the total fused signal and the output of each sensor separately. The simulation results showed that the proposed approach using FCM and fusion engine based COG techniques improved the diagnosis accuracy.