Convolutional Neural Network Approach for Aircraft Noise Detection

Abstract

People living near the airport are experiencing many inconveniences due to frequent aircraft noise. For these people, the government uses the aircraft noise evaluation unit (e.g., Lden) to calculate the degree of annoyance and then compensate for aircraft noise. Aircraft noise evaluation unit should be calculated only by aircraft noise, but the reality is not so. This is because the aircraft noise monitor measures not only aircraft noise but also loud background noise. Therefore, in this paper, we propose a method of recognizing only the aircraft noise among the stored noise from the noise monitor to calculate accurate aircraft noise evaluation unit. The proposal uses convolutional neural network, one of the deep learning techniques. Our proposal purposes less than 1% false-positive (FP) or false-negative (FN) rate.