Prediction Of The Noise Of Ship's Superstructure Cabin Based On SVM Method

In this paper the noise of superstructure cabins of large container ship and the noise of superstructure cabins of bulk carrier was respectively predicted, basing on nonlinear regression of support vector machine (SVM).The prevalent methods used nowadays in the filed of ship noise prediction are discussed, and then the method nonlinear regression in SVM is applied to predict the cabin noise of ship superstructure for the first time. The basic knowledge of acoustics, main sources of ship noise, the transmit route and characteristic were represented, aiming to provide the theoretical knowledge for the following noise prediction. Through the analysis of the actual measured superstructure noise of the 174000DWT bulk carrier, the noise feature and the main noise sources of the ship were obtained, aiming to provide the basic reasons for the selection of the main noise sources. The nonlinear regression of SVM was programmed by Matlab codes, which is verified by the examples in references. Then two nonlinear regression SVM models of the superstructure cabin noise of large container ship were established, taking the measured noise data of the superstructure cabins of one container ship as the training sample. By comparing the results of the two models, the better model in application was used to predict the superstructure cabin noise of the 5618 TEU, 4100 TEU and 17400DWT bulk carrier. Then the method feasibility was certified by comparing and analyzing the predicted data and measured data.