Electromagnetic and acoustic emissions to diagnose complex electrical and mechanical structures

A diagnostics procedure based on signal processing of electromagnetic and acoustic emissions aimed at investigating complex electrical and mechanical structures is described. Acquired signals emitted by a structure can be activated by either natural or artificial stimuli; when using artificial and known stimuli, that is to say, stimuli linked to a set of parameters representing "state vectors", the diagnostic method may prove to be very accurate. Signals are processed through the following stages: acquisition, segmentation, parameterization, classification and recognition. Many complex structures were examined, and some relevant real diagnostics cases concerning power transformers and power circuit breakers, are illustrated. Finally, an original implementation of the wavelet technique for the analysis of the acoustic sampled signals is illustrated.