Development of a portable NDE system with advanced signal processing and machine learning for health condition diagnosis of in-service timber utility poles

Aiming at current shortcomings of non-destructive evaluation (NDE) in health condition esti-mation of timber utility poles, this paper put forward a novel testing method via combination of a portable NDE system, advanced signal processing and machine learning techniques. Primarily, the multi-sensing strat-egy is employed and incorporated in current NDE technique to capture reflected stress wave signals, avoiding difficult interpretation of complicated wave propagation by only one sensor. Secondly, advanced signal pro-cessing methods, such as ensemble empirical mode decomposition (EEMD) and principal component analysis (PCA), are introduced to extract effective wave patterns that are sensitive to structural damage. Moreover, based on captured signal features, the state-of-the-art machine learning techniques are applied to implement the condition assessment. Finally, field testing results of 26 decommissioned timber poles at Mason Park in Sydney are used to validate the effectiveness of the proposed method.