Abstract
A monitoring method using acoustic diagnosis technique is proposed to improve the management of shield tunneling operation. Relative energy contained in a specific frequency band in power spectrum to the whole energy is used for detecting anomalies in the machine health. The investigation on data sets obtained from two construction sites shows that the signals associated with the anomalies have an energy ratio over 0.4. Moreover, principal component analysis is employed to perform the pattern classification of the signals for detecting the changes in ground conditions. The study shows that the normalized power spectrum is useful for the classification.
