Wavelet transform based sensor validation

This paper describes a novel approach based upon wavelet transforms for sensor validation. By the use of wavelet transforms that well localise the characterisation of a signal both in time and frequency domains, the occurring instants of abnormal status of a sensor in its output signal can be identified. Once these instants are detected, the distributions of signal energy, on all decomposed wavelet levels of this signal before and after these instants, are used to validate the status of the sensor because there are various energy distribution patterns for different failures of a sensor. Both computer simulation and experimental data from a general-purpose pressure transducer have verified the effectiveness of the proposed method.