Research On Method Of Aircraft Noise Recognition

With the rapid development of the civil aviation industry, the number of new airports isincreasing, it causes an increasingly serious noise pollution problem around the airport,Airport noise monitoring system is an important mean of effective monitoring andmanagement of the noise pollution. An essential partment in the airport noise monitoringsystem contains many noise monitoring equipments widely emplaced in various regionsaround the airport, especially in residential areas, this inevitably leads that those noise datacollected by these noise monitoring equipment may not be just the aircraft noise we careabout, and may also include noise come from cars, motorcycles and other ambient which willcause great difficulties to subsequent data processing, an effective aircraft noise recognitionalgorithm is required.With the help of theoretical study and experimental verification, one more suitabletheoretical method for identification of aircraft noise is proposed. Based on analysis of thephysical characteristics of the aircraft noise, the framework of the identification of the aircraftnoise is constructed, the noise reduction effect of aircraft noise the wavelet and waveletpacket is compared by conducting several experiments, experimental results show that thewavelet packet noise reduction method can preserve more high-frequency feature information,and so it is very suitable for the acoustic characteristics of the aircraft noise. Two acousticfeature extraction algorithms MFCC and perceptual linear predictive PLP are improved andintegrated by adding a feature-enhancement step to improve recognition performance underconditions of low SNR of the algorithm. In the final stage,SVM is chosen as the finalclassifier,as its kernel function parameter and the penalty factor is very important for theperformance of the final classifier, so an improved genetic algorithm of SVMparametersoptimization is applied to determine the best values of the parameters. Finally,experiment results verify the effectiveness of the design of aircraft noise recognitionalgorithm.