A CNN-Based Method for Enhancing Boring Vibration with Time-Domain Convolution-Augmented Transformer

SIMPLE SUMMARY: Trunk-boring insects have emerged as one of the most threatening forest pests globally, causing significant damage to forests. Certain groups of larvae reside within tree trunks without any observable external signs indicating their presence. This poses a significant challenge for pest detection. To tackle this issue, acoustic techniques are frequently utilized, which involve embedding a vibration probe into the tree trunk to capture the vibrations produced by larvae and using a model to distinguish whether the tree is infested. However, this approach requires the acquisition of the purest possible vibrations signal. Thus, to ensure accurate analysis, a noise suppression process is crucial, since the signals collected in real-world environments are often subjected to varying degrees of environmental noise interference. In this study, we employed artificial intelligence techniques to develop a boring vibration enhancement model. The training data utilized in this study comprise boring vibrations captured from trunk sections and typical environmental noise present in the trees’ habitat. The experimental results demonstrate that the enhancement method proposed by our model significantly boosts the performance of an established classification model. Overall, this study promotes the development of sustainable and efficient forestry protection approaches by improving the accuracy of pest detection. ABSTRACT: Recording vibration signals induced by larvae activity in the trunk has proven to be an efficient method for detecting trunk-boring insects. However, the accuracy of the detection is often limited because the signals collected in real-world environments are heavily disrupted by environmental noises. To deal with this problem, we propose a deep-learning-based model that enhances trunk-boring vibration signals, incorporating an attention mechanism to optimize its performance. The training data utilized in this research consist of the boring vibrations of Agrilus planipennis larvae recorded within trunk sections, as well as various environmental noises that are typical of the natural habitats of trees. We mixed them at different signal-to-noise ratios (SNRs) to simulate the realistically collected sounds. The SNR of the enhanced boring vibrations can reach up to 17.84 dB after being enhanced by our model, and this model can restore the details of the vibration signals remarkably. Consequently, our model’s enhancement procedure led to a significant increase in accuracy for VGG16, a commonly used classification model. All results demonstrate the effectiveness of our approach for enhancing the detection of larvae using boring vibration signals.