Short-term motion prediction of a semi-submersible by combining LSTM neural network and different signal decomposition methods

Abstract

Reliable short-term motion prediction can improve the safety of marine operations. In order to improve the prediction accuracy, the original motion signal is often decomposed into several uncoupled sub-series by using the empirical mode decomposition (EMD) technology. Based on the traditional EMD method, several new signal decomposition methods have been proposed, such as ensemble empirical mode decomposition (EEMD), complete ensemble empirical mode decomposition with adaptive noise (CEEMDAN) and empirical wavelet transform (EWT). However, there is still a lack of detailed comparative study research on the influence of different signal decomposition methods on the prediction accuracy of short-term motion. In this study, several hybrid prediction models for short-term motion of semi-submersible are established by combining different signal decomposition methods and the LSTM model, namely EMD-LSTM, EEMD-LSTM, CEEMDAN-LSTM and EWT-LSTM models. Through the comparative analysis of the prediction results of different models, the influence of different signal decomposition methods on the prediction accuracy of short-term motion is investigated. It is found that the EWT-LSTM model consumes the least computing resource, but presents the best performance among different hybrid models.