Improving Engine Sound Analysis in Hot-Test Environments via a RAB-U-Net (Residual Attention Block U-Net) Noise Removal Method

The study introduces a Residual Attention Block U-Net (RAB-U-Net) for enhancing engine sound analysis by effectively removing background noise during hot tests on production lines. This deep learning model improves the accuracy of engine noise detection compared to traditional methods, demonstrating its potential for real-time applications in automotive diagnostics. The advancement is significant for practitioners as it leverages neural network architectures to enhance sound analysis, thereby improving product quality and performance assessments in manufacturing environments.