Predicting Performance of Unattended Machinery Plant: A Step Toward Trustworthy Autonomous Shipping

EasyChair Preprint 4219

Abstract

Future Maritime Autonomous Surface Ships (MASS) are expected to have considerable impact on maritime trade. For development of autonomous operations, it is vital a ship’s ability to monitor system health, establish and prognosis what is going wrong around it and make decisions based on that information. Many tasks that are currently done by a ship’s crew are not so easily replaced by a machine in unmanned vessel. This complicates maintenance and repair of broken equipment when crews are intending to be removed in autonomous shipping. By the absence of on-board crews, the operation will be susceptible to emerging risks and will have a big impact on the design of the system such as the engine rooms. Further to this, the existing market and regulatory arrangements mainly focused on investigating how advanced control systems, navigation software and online communications could control an unmanned vessel. What is often neglected is the importance of trustworthy and enabling effective maintenance for the unattended system integrated with autonomous ships. This paper proposes a dynamic predicting approach for the reliability of the power plant when it is unmanned. A systematic framework is presented for modelling failures propagated through the machinery that result in system disruption. The main concern is to understand the failure events that will put the system performance at risk without human intervention. To this end, a “closed-loop” reliability model is constructed to act as a framework for evaluating the degradation of system under the influence of disruptive event. A real case study of Maine Engine (ME) is adopted to demonstrate the application of the presented research framework. The results highlight the importance of predicting appropriate hazard rate functions that is essential for redesign of unattended machinery according to the maintenance activities.

Keyphrases: Autonomous Shipping, Bayesian inference, Multinomial Process Tree, maritime autonomous surface ship, reliability engineering, system reliability, unattended machinery plant

@booklet{EasyChair:4219,
  author    = {Mohammad Mahdi Abaei and Ahmad Bahootoroody and Ehsan Arzaghi},
  title     = {Predicting Performance of Unattended Machinery Plant: A Step Toward Trustworthy Autonomous Shipping},
  howpublished = {EasyChair Preprint 4219},
  year      = {EasyChair, 2020}}