Prediction interval

Statistics

Operating system

Algorithm

Combinatorics

Machine learning

Physics

Artificial intelligence

Nonlinear system

Gaussian

Geometry

Interval (graph theory)

Point (geometry)

Mathematics

Computer science

Artificial neural network

Radar

Gaussian process

Term (time)

Quantum mechanics

Ground-penetrating radar

Telecommunications

Kriging

Regression

Process (computing)

Applied Ocean Research

The random motions of a ship due to the sea waves directly affects the safety and efficiency of marine operations. Accurate short-term ship attitude prediction plays a vital role in operational decision-making in future seconds. In this paper, a new hybrid prediction model of ship motion attitude is proposed based on Long Short-Term Memory (LSTM) neural network and Gaussian Process Regression (GPR). When dealing with nonlinear regression problems, LSTM model can get high-accuracy point prediction results, while GPR model with lower prediction accuracy can obtain interval prediction results with probability distribution significance. The LSTM-GPR model successfully combines the advantages of LSTM and GPR, and can obtain high-accuracy point prediction results and reliable interval prediction results at the same time. The prediction experiments of ship rolling angle and pitch angle are carried out under both motion and static conditions. The results show that the LSTM-GPR hybrid model can obtain reliable interval prediction results without reducing the forecasting accuracy of LSTM model, which verifies the effectiveness and advancement of the hybrid model.

Short-term ship motion attitude prediction based on LSTM and GPR