Using enhanced cepstral analysis for structural health monitoring The thesis is based on damage detection algorithm applied to a real case study (masonry buildings). From artificial input, training Time Histories of the undamaged Model are created; then, the structure is artificially damaged with the reduction of the E modulus of the material of some sections, and new Time Histories are produced. These last ones will be the testing samples for the damage detection algorithm. The main study is conducted on the Filterbanks for the preprocessing of the signal.