abstract
- Fault detection strategies are important for ensuring the safe and reliable operation of mechanical and electrical systems. Recently, a new signal-based fault detection and classification strategy has been proposed, which makes use of artificial neural networks (NNs) and the smooth variable structure filter (SVSF). The strategy, referred to as the NN-SVSF, has shown promising results with applications to benchmark classification problems. New developments of the SVSF have resulted in improved performance in terms of state and parameter estimation. These developments are used to enhance the NN-SVSF in an effort to further advance the signal-based strategy. This paper studies and compares the results of applying other popular strategies on an internal combustion engine (ICE), for the purposes of fault detection and classification. © 2012 IEEE.