Abstract: The noise induced by the high-speed trains has always been one of the main problems restricting the development of new-type trains. In particular, low frequency noise has been the focus of researchers due to its characteristics such as low attenuation, long propagation distance and strong permeability. Due to its non-contact and high-speed characteristics, the main noise source of maglev train is changed from the mechanical noise between wheels and track to the aerodynamic noise caused by pulsating air pressure. The noise level mainly depends on the near-field aerodynamic characteristics. When the speed increases, the influence of the near-field aerodynamic characteristics on the interior noise is particularly severe. In order to solve the noise problem, the study of the near-field aerodynamic characteristics of high-speed maglev trains had became a key step in the research. When the speed increased, the influence of the near-field aerodynamic characteristics on the interior noise was particularly severe. In order to solve the noise problem, the study of the near-field aerodynamic characteristics of high-speed maglev trains had became a key step in the research. Therefore, based on the Lighthill acoustic theory, this study adopted the FW-H acoustic model for a certain type of three-group high-speed maglev train, and used the large eddy simulation method to solve the surface turbulence of the high-speed train. The simulation analyzed the steady-state aerodynamic characteristics and transient aerodynamic noise excitation source characteristics of the 1:10 scale model of the train with a speed of 600 km/h under the open-wire condition without considering the track factor. The results showed that the pressure changes were correspondingly obvious at the positions with large curvature changes on the surface of the high-speed maglev train. At the same time, the positions with large curvature changes also made the motion of the flow field eddy current in the near-wall region more intense. The pressure gradient was large and the pressure value at the nose tip of the car was significantly higher than that of the nose tip of the car rear. The sound power level of the train surface can reach up to 155 dB, and the main noise sources were generated in the nose tip area of the car, where the surface curvature changed greatly, and the connection between the carriages. In the low frequency band, the distribution law of the sound pressure level at the monitoring points on the front and rear surfaces of the vehicle was the same, showing the phenomenon that the sound pressure level decreased with the increase of the frequency, and the overall sound pressure level of the monitoring point at the nose tip of the vehicle was the same higher than the nose of the car. The sound pressure level distribution law of the two monitoring points at the transition position of the front and rear of the vehicle to the body was similar, with obvious fluctuations in the low frequency band around 500 Hz, and then the sound pressure level gradually decreased with the increase of the frequency. The change rule of the sound pressure level at a speed of 500 kilometers per hour was the same as that at 600 kilometers per hour. The overall sound pressure level of the monitoring point was lower than 600 kilometers per hour, and the change of the monitoring point at the rear of the vehicle was the most obvious. The research results obtained in this paper can provide a certain scientific basis and guidance for the calculation of in-vehicle noise of high-speed maglev trains and the subsequent optimization of low-frequency noise, so as to meet passengers' requirements for ride comfort.