Mechanical structures suffer from degradation due to alternative stresses, corrossion, radiation... Some of them might be assumed to have unlimited life, others need to be overhauled so the properties are restored for a full service use. Nowadays, an extensive study and/or simulations could be made in order to asses with a certain degree of confidence and safety factor, what the overhaul periods shall be. There is a huge cost improvement though, when the integrity is measured in real-time and the maintenance is precisely done once the damage is detected or about to be, before it is irrevesibly damaged or results in catastrophic consequences. Structural Health Monitoring (SHM) is a set of methods for this purpose. The physical phenomena it bases on, is the propagation model of the acoustic waves on the volume of the structure. In the current work, this propagation model is extracted using Information Theory methods by means of Principal Component Analysis and Independent Components Analysis. This model is extracted from a healthy structure. A test data from different experiments is compared with this model and further statistical analysis is performed for assessing how far from the model we are, assuming that a certain distance from the model means a damaged structure. The present work takes on the underpinning Structural Health Monitoring state-of-the-art methods and focuses on improving the sensitivity of detection and the robustness of the constructive part of the acquisition system.