Timber Health Monitoring using piezoelectric sensor and machine learning

Abstract

The Timber Health Monitoring System, which enables constant monitoring of wooden buildings by artificial intelligence based analysis of the signals of a piezoelectric sensor attached to a piece of timber, is proposed. Basic verification was carried out by modeling timber damage and performing vibration tests. Analysis of the obtained waveform data using the k-nearest neighbor (k-NN) method and a support vector machine revealed that the proposed system has a strong classification performance. We also tried reducing the data dimensions by using principal component analysis and found that the classification rates barely decreased even if dimensional reduction was adopted. These results are promising for the realization of our proposed system.