Study on Wave Impact Force Prediction of Different Shore Connecting Structure Based on Improved BP Neural Network

In this paper, the advanced neural network technology was introduced to the field of the wave impact force prediction. A three-layered BP neural network is employed and the units of input layer are wave style, wave period, incident wave height, relative clearance, relative width of wharf deck, and reflection coefficient of different shore connecting structure. The unit of output layer is the incident wave impact force. Compared with traditional BP algorithm and some other improved algorithm, It is shown that Levenberg-Marquardt Algorithm is more effective on the prediction of wave impact force.