Aircraft class recognition based on dynamic hierarchical weighting of multiple neural networks outputs

Aircraft noise is a major concern for current world-wide airports. Evaluation of airport noise pollution mainly depends on the correlation between the aircraft class, the noise measured and the flight path. Certification, evaluation and regulation procedures usually require the foregoing correlation to be performed by means of different sources of information beyond that provided by the aircraft itself. In this regard, methods to identify the aircraft class taking off based on features extraction from the noise signal have been developed. This paper introduces a new model for aircraft class recognition based on signal segmentation and dynamic hierarchical weighting of K parallel neural networks outputs O_p^k. Performance of new model is benchmarked against models in literature over a database containing real-world take-off noise measurements using three different features types. The new model is more accurate regarding the above mentioned database and successfully classifies 87% of measurements.