Motion and Load Prediction of Floating Platform in South China Sea Using Deep Learning and Prototype Monitoring Information

Abstract

The offshore floating platform shows the strong nonlinear characteristics subject to harsh ocean environmental conditions. It is of practical significance and engineering value to predict the marine environmental load and platform motion response by using the prototype monitoring information. Meanwhile, under the interaction of the high-frequency six degrees of freedom (DOF) movements of the platform. Based on the long-term prototype monitoring data of a semi-submersible platform in the South China Sea, the present paper mainly studies the following two aspects: 1.The prediction of ocean environmental load considering time correlation is studied based on the method of long-short-term memory (LSTM) neural network with the combination of the field monitoring data. The comparison between predicted and measured results shows that the present prediction method has the high accuracy and low computation cost. Besides, this method can be extended to short-term predictions of other environmental loads. 2.The nonlinear mapping relationship between the ocean environment load and the floater motions is constructed based on the deep learning method. The simulated results indicated that the mapping relationship can be used to predict the six DOFs motions of the platform with high accuracy by using the forecasting prototype monitoring data and ocean weather information. Based on this research and the short-term prediction of environmental loads, we can do some studies on short-term prediction of floater motions in the future.