Numerical optimization of flow noises for mufflers based on the improved BP neural network

Aimed at the large noise of tail pipe, the method of fluid dynamics was firstly applied to analyze the inner flow field of the exhaust muffler. According to the result, the large noise of tail pipe was mainly caused by air flow regeneration noise, and the vice muffler was not the major component for generating airflow noise. The largest pressure of the whole muffler system was at the outlet end of main mufflers. The largest flow velocity was in the connection pipe between main mufflers and vice mufflers. Secondly, boundary element model of transmission loss for the muffler was established to compare and analyze it with the experimental. The experimental and computational value of transmission loss for the muffler has a good consistency in both change trend and numerical value, and the computational model was reliable. Finally, GA-BP neural network algorithm was used to optimize the acoustic performance of the muffler. Airflow noises of the tail pipe were effectively reduced through optimizing the inner structure of the muffler.