Research on engine exhaust noise control based on the neural network

Abstract

A new type of semi-active control method which called bypass duct silencer is proposed. For the sake of reducing the exhaust noise by 15 dB, the relationship among the engine speed, the exhaust temperature and the structure of the bypass duct silencer is got. Then the bypass duct silencer which can change the internal structure of the silencer is present. The silencer contains several valves, which could control the flowing direction of sound wave aiming to neutralize sound wave in the downstream intersection of the duct. When the BP neural network structure was trained, the engine speed and the exhaust temperature measured were considered as input signals, meanwhile, the controlling situation of the valves was considered as output signal. The network structure is certain when the error of this method lies within a tolerable range the thresholds and weights are determined. The trained neural network structure could be used to reduce the exhaust noise in practical application.