Research on four-layer back propagation neural network for the computation of ship resistance
Aiguo Chen, Jiawei Ye · 2009
Applying the original experimental data of series 60 ship models, four-layer back propagation neural network is founded. Test samples and interpolated samples are randomly selected as input vectors. The worse of the maximum relative error, the average relative error and the correlation coefficient between the outputs and the goals, their regression lines and the performance curves plotted by the neural network are used to measure the performance of the network. By data pretreatment manner, making a great deal of experiments to optimize the training functions, the performance functions, the transfer functions and the neuron number of latent layers, develop the optimal four-layer back propagation neural network for the computation of ship resistance. Easily and quickly calculating ship resistance, the neural network can be applied to research the performance of ship resistance, the optimization of hull form and the optimal matching design of ship engine and propeller.