Ship tail gas detection precision is an important basis for emission factors and ship pollution detection.Due to the influence of fuel quality, water content and oxygen saturation, fuel combustion,diesel engine unit and the actual operating conditions of the purification equipment, the emission of the exhaust gas fluctuates greatly. In order to make up for all kinds of Interference factor of the detection accuracy, an improved BP neural network which in data fusion algorithms be accepted. On the basis of real data which NOX emission under actual working conditions of inland waterway ships, the computer simulation and analysis show that this method can improve the accuracy effectively.