With the widespread adoption of electric vehicles, reducing aerodynamic noise from exterior components, such as side-view mirrors, has become critical for improving cabin comfort. Consequently, identifying the sources of such noise is essential for effective design optimization [1]. However, source identification via iterative forward Computational Fluid Dynamics (CFD) simulations requires prohibitively large computational resources. Alternatively, traditional inverse analysis methods, such as those based on adjoint equations, are theoretically challenging to formulate and often suffer from numerical instability due to the nature of backward-in-time integration. To overcome these limitations, this study proposes an inverse analysis framework based on Physics-Informed Neural Networks (PINNs) [2]. Unlike purely data-driven methods, PINNs directly incorporate governing physical laws, specifically fluid and acoustic equations, into the loss function. This approach reformulates the inverse problem as a robust optimization task, thereby avoiding the complex derivation of adjoint equations and ensuring numerical stability. In this study, we adopt a PINN framework that uses pressure field data obtained from high-quality numerical simulations to estimate the locations of aeroacoustic sources. The proposed method is investigated through two numerical examples: a two-dimensional problem with an analytical solution and a three-dimensional problem involving a realistic automotive side-mirror geometry. Through these examples, we discuss the potential of the PINN-based approach to identify aeroacoustic source information from pressure field data. This framework is expected to provide a promising approach for efficient and accurate noise source identification and contribute to the development of quieter vehicles. [1] A. Tajima, J. Ikeda, K. Nakasato, T. Kamiwaki, J. Wakamatsu, M. Oshima, C. Li, and M. Tsubokura, Numerical Simulation of Fluctuating Wind Noise of a Vehicle in Reproduced on-Road Wind Condition, SAE International Journal of Advances and Current Practices in Mobility, Vol. 7, pp.240-253, 2025. [2] M. Raissi, P. Perdikaris, and G. E. Karniadakis, Physics-Informed Neural Networks: A Deep Learning Framework for Solving Forward and Inverse Problems Involving Nonlinear Partial Differential Equations, Journal of Conputational Physics, Vol. 378, pp.686-707, 2019.