Recently, Associate Professor JIN Yao from the School of Building Services Science and Engineering has made two landmark contributions to the mathematical modeling of aerodynamic noise. The corresponding studies have been successively published in Applied Mathematical Modelling, a prestigious interdisciplinary journal in applied mathematics (SCI Q1 Top Journal, Chinese Academy of Sciences SCI Zone 1). Associate Professor JIN Yao is the sole first author of both papers, and XAUAT is the first affiliated institution for both publications.

The first paper, entitled "Assessment of recent empirical wall-pressure auto-spectrum models under various pressure gradient conditions", presents a systematic evaluation of 12 representative empirical models for wall-pressure auto-spectrum developed in recent years. It examines the effects of Reynolds number (Re) and non-equilibrium pressure gradients (dp/dx) on the scaling laws of energy spectra, and clearly reveals the deficiencies of existing models in describing the scaling laws for low-frequency spectral uplift and high-frequency decay. Based on these findings, a frequency-adaptive weighted spectral modeling strategy is proposed, which is capable of simultaneously capture the scaling behavior over the low-, medium-, and high-frequency bands. In addition, an analysis of the physical consistency of parameters in emerging intelligent regression models, including those based on Gene Expression Programming (GEP), show that such models fail to reproduce the correct physical trends and therefore do not yet possess reliable physical interpretability.

Figure 1: The frequency-adaptive weighted model shows better agreement with the experimental data in both the low- and high- frequency scaling laws.

The second paper, entitled "Data-driven modeling for rapid prediction of aerodynamic noise directivity of flow over a cylinder", develops three noise directivity prediction models that integrate data fitting techniques with acoustic radiation theory. These include: a) a fully data-driven Proper Orthogonal Decomposition (POD) model based on Singular Value Decomposition (SVD) and interpolation-based reconstruction, b) a semi-analytical model that takes fluctuating aerodynamic forces on the body surface as input, and c) a semi-analytical model based on reconstructed coefficients of the velocity potential function. The results show that all three models achieve high prediction accuracy with extremely low computational cost, thereby providing a novel and efficient framework for the rapid prediction of aerodynamic noise directivity across a range of operating conditions.

Figure 2: Comparison of acoustic directivity results from direct noise simulation, the force-based model, the dipole-based model and the POD-based model.

Paper Links:

//www.sciencedirect.com/science/article/pii/S0307904X2500229X

//www.sciencedirect.com/science/article/pii/S0307904X25000125