An intelligent thrust reverse noise detector is presented in this paper (TREND). A first detector is customized for the detection of landing sound events. When one of those events is detected, a second detector is triggered to detect thrust reverse. In the case that both events are detected, each of them have to be classified in two separate block. If the first events is classified as landing, and the second one as thrust reverse, the system identifies the activation of thrust reverse. The detection is based on thresholds applied to the sound power level (instead of sound pressure level in the traditional approaches) time histories, which are estimated using a microphone array and an inverse sound propagation model. This approach has worked well, as the estimation of sound power level enhances the sound events and their separation, even if landing and thrust reverse are close to each other. The classification is implemented through pattern recognition techniques, which reduces the number of false positive in the detection stages. The results obtained in Madrid-Barajas airport through the implementation of the methodology in this TREND tool are promising with error rates lower than 10%.
展开▼
