Pattern recognition adaptive dynamic neural network model for aircraft noise generated in airport urbanized area

For some time, in the DICAAr(1), it was deepened a specific research theme on the different issues related to the impact of air and noise type on adjacent areas to major airports, due to air traffic overflying. In particular a INM6 simulation model, able to produce, through the identification of the noise contours, an acoustic zoning, covering both airport surfaces and those adjacent thereto has been calibrated. A more recent deepening of the topic concerned, however, the identification of a neural network for the recognition of the noise produced by airplanes, always during the LTO cycle; its findings are reported in this paper. This procedure, however, is excessively complex on the implementation plan as linked to a continuous experimental comparison between the sound recording units positioned in the vicinity of the airport and the airport newspaper supplied to the air traffic control bodies. A neural network model to recognize the noise generated by airliners has been considered. A certain number of experimental measurements has been performed in order to implement the training and validation set of the neural model. The experimental measurements include the noise of several type of aircraft engines during the take off and the landing operating mode. The aim of the study is to develop a model able to predict the aircraft noise under different environment and climate conditions. The results in terms of ability to recognize the exact aircraft noise even if not belonging to the training set or under not standard enviroment and climate conditions will be discussed