Abstract: in order to decrease the cost though predicting the noise before building to improve the ship. This paper introduces the prediction of the noise of the ship’s superstructure cabins based on SVM method. This paper establishes two nonlinear regression models of the noise of the superstructure cabins of large container ship, which is based on Support Vector Machine, using RBF kernel function and ERBF kernel function, taking the actual measurement noise of one container ship’s superstructure cabins as the training sample. The two models are applied to predict the noise of the parent ship and another container ship’s superstructure cabins. Compare to the result and choose the better model, then predict the noise of 174000DWT bulk carrier’s superstructure cabins. The result shows that, it is feasible to predict the noise of the ship’s superstructure cabins，and the effect is good.