Dynamic Windowing Algorithm to Improve Ship Response Prediction in Transitory Conditions

8 pages•Published: January 5, 2024

Abstract

The emerging trend of autonomous shipping has demanded the automation of many onboard systems and sub-systems to minimize human involvement in decision-making. Given the time- varying nature of the sea, the respective knowledge of ship response possesses a variety of applications in real-time operations and guidance, where predicting responses a few seconds ahead plays a crucial role in dynamic control. Within current literature, many long-run studies lack evaluation of the dynamics of ship response time-series in a transitory condition between sea states. To address this deficit, the present study proposes an innovative algorithm to automatically detect the substantial changes in the time series regime for ship response data and adjust the observation window for prediction accordingly. For prediction purposes, a well-known classic nonlinear regressor, Seasonal Auto-Regressive Integrating Moving Average (SARIMA) has been employed in an adaptive sense. Despite this concept being primarily developed for advancing ship maneuvering, experimental and simulation data of a moored semi-submersible vessel has been utilized, given the dynamics' simplicity. The results demonstrated the efficiency of the proposed filter in minimizing uncertainty in ship response prediction according to the prevailing sea conditions. Although the proposed algorithm shortens the prediction length in transitory signals, it essentially improves the prediction results, for the estimation models are built only on informative short-term data. The proposed workflow can not only increase the autonomy of the involved system with ship response data, but also be further used on any onboard system dealing with time-varying information.

Keyphrases: Autonomous ships, Intelligent Systems, Seakeeping, short-term prediction, time series analysis

In: G. Reza Emad and Aditi Kataria (editors). Proceedings of the International Conference on Maritime Autonomy and Remote Navigation 2023, vol 2, pages 59--66

Links:	https://easychair.org/publications/paper/HNBr
	https://doi.org/10.29007/2gnp