Improved ellipsoid and nonlinear function model for aircraft noise prediction

Abstract

Existing aircraft noise models often rely on static databases and fail to comprehensively account for dynamic environmental and operational factors, resulting in relatively significant prediction errors in a single flight noise event. To make up these issues, a noise attenuation model integrating ellipsoid and nonlinear function has been proposed for the first time, this paper will further improve our previously proposed model. The key improvements are two aspects: firstly, geometric constraints derived from the analysis of NPD database to combined to ensure that the noise attenuation curve remains physically reasonable, overcoming the limitations of previously pure data fitting. Secondly, the objective function is redefined, thereby reducing the errors that assuming a constant sound source level in the original model, and improving the model’s balance and extrapolation performance. The comparison results show that the ME, MAE and fluctuation range of the calibration and validation errors have all been reduced to some extent. The improved model also obtains a more balanced error distribution in predicting the EPNL. The sound level attenuation curves generated are more concentrated, and closer to the noise attenuation process in real environments. It is more conducive to analyze the noise attenuation characteristics of aircraft noise and establish corresponding noise databases. Furthermore, the model simulates the directivity of noise propagation through a simple geometric parameter adjustment of ellipsoid model, demonstrating a good environmental adaptability. The value of these enhancements not only lies in combining physical laws with data fitting, further enhancing the performance of the model, but also offering a new research direction for developing a more adaptable, interpretable, and reliable aircraft noise model.