Standing tree health assessment using contact-ultrasonic testing and machine learning

The problem of hole-defect detection in standing trees is solved. An ultrasonic device (Pundit PL-200) was employed to collect ultrasonic signals from various wood specimens both in the lab and field. The collected ultrasonic signals were then processed through the Variational Mode Decomposition algorithm to derive effective features. In order to solve the classification problem more efficiently, the obtained characteristics were then analyzed through PCA to determine the most useful features. Several machine learning algorithms and a one-dimensional convolutional neural network (1D-CNN) were employed to solve a set of classification problems based on data collected from (1) specimens with artificial defects in the lab and (2) billets with natural defects selected from trees harvested at sites in the two states of WA and NSW, Australia. The results demonstrate the effectiveness of the proposed method for classifying wood materials based on their health state, where testing accuracy results of 100% in the lab and at least 92.2% in fields were achieved. The Fine Gaussian SVM was found to perform most effectively on data derived from specimens in the lab and fields. It was also shown that 1D-CNN results were more reliable for generalizing the solution to the classification problem of standing trees in fields.

Automated summary

The problem of hole-defect detection in standing trees is successfully solved using ultrasonic signals and machine learning algorithms. The proposed method, which includes Variational Mode Decomposition and PCA, achieves high accuracy in classifying wood materials based on their health state. A one-dimensional convolutional neural network (1D-CNN) is also employed to generalize the solution to the classification problem of standing trees in fields.