Inadequate ship machinery maintenance can increase equipment failure posing a threat to the environment, affecting performance, having a great impact in terms of business losses by reducing ship availability, increasing downtime and moreover increasing the potential of major accidents occurring and endangering lives on-board. This paper aims to provide a systematic approach for identifying critical ship machinery systems/components and to analyse their physical parameters. Critical ship main engine systems/components are used as input in a dynamic time series neural network, in order to monitor and predict future values of physical parameters related to ship critical systems. The critical main engine systems/components and their relevant parameters to be monitored are identified though the combination of Fault Tree Analysis (FTA) and Failure Mode and Effects Analysis (FMEA). A case study of a Panamax size container ship is presented in which Artificial Neural Networks (ANN) are used to predict the upcoming values of all main engine cylinders exhaust gas temperatures. The forecasted results were validated through comparison with actual observations recorded on board the ship. The proposed hybrid methodology successfully presents a systematic approach for initially identifying critical systems/components through reliability modelling and tools and subsequently monitoring their physical parameters through neural networks. 2. Research Background/Literature The evolution of maintenance was based not solely on technical but rather on techno-economic considerations according to Pintelon and Parodi-Herz (2008). Furthermore, according to Arunraj and Maiti (2007), maintenance policies can be categorised into four generations as seen in Figure 1. The fourth generation is the most recent one, which focuses on condition based maintenance, condition monitoring and failure eliminations. It concentrates on reducing the proportion of equipment failures and overall levels of failure probability through various tools and strategies, based on preventive and predictive maintenance approaches.

Condition monitoring is the process of monitoring parameters expressing machinery condition, interpreting them for the identification of change which could be indicative of developing faults. Data pre-processing and post-processing is of great importance in a ship condition monitoring software tool, as misinterpretation of data can significantly affect the accuracy and performance of the predictions made. In this paper, data for key physical performance parameters for a PANAMAX container ship main engine cylinder are pre-processed and clustered using a two-stage approach. Initially, the data is clustered using the ANN (Artificial Neural Network)-Self-Organizing Map (SOM) and then the clusters created by the SOM are interclustered using the Euclidean distance metric into groups. A custom algorithm using a combination of logical operators and conditional statements is used to compare cluster distances and obtain neighbour clusters containing similar data. The case study results demonstrate the capability of the SOM to monitor the main engine condition by identifying clusters containing data which are diverse compared to data representing normal engine operating conditions. The results obtained from the clustering process can be further expanded for application in diagnostic purposes, identifying faults, their causes and effects to the main engine of a ship.

Though the maritime industry is still predominantly reliant on a time-based, prescriptive approach to maintenance, the increasing complexity of shipboard systems, heightened expectation and competitive requirements as to ship availability and efficiency and the influence of the data revolution on vessel operations, favour a properly structured Condition Based Maintenance (CBM) regime. In this respect Artificial Neural Networks (ANNs) can be applied for predictive maintenance strategies that can assist decision makers in selecting appropriate maintenance actions for critical ship machinery. This paper focuses on developing a Nonlinear Autoregressive with Exogenous Input (NARX) ANN model for forecasting future values of performance parameters of a marine main engine. Moreover, a detailed sensitivity analysis is conducted to examine the performance and robustness of the developed NARX model, based on the dataset applied for its training. Through the NARX model, a predictive monitoring approach can be achieved for ship machinery monitoring as high forecasting accuracy was achieved for the case study of a main engine cylinder exhaust gas outlet temperature. The sensitivity analysis overall demonstrated the good performance and generalisation of the NARX model as it successfully considers different values of the time series data for conducting the onestep-ahead output.

The proposed research, through INCASS (Inspection Capabilities for Enhanced Ship Safety) FP7 EU funded research project tackles the issue of predictive ship machinery inspection by enhancing reliability and safety, avoiding accidents, and protecting the environment. This paper presents the development of Machinery Risk/Reliability Analysis (MRA). The innovation of this model is the consideration and assessment of components' risk of failure and reliability degradation by utilizing raw input data. MRA takes into account the system's dynamic state change, concerning failure rate variation over time. The presented methodology involves the generation of Markov Chains integrated with the advantages of Bayesian Belief Networks (BBNs). INCASS project developed a measurement campaign, where real time sensor data is recorded onboard a tanker, bulk carrier and container ship. The gathered data is utilized for MRA DSS tool validation and testing. Following research involves components and systems interdependencies and feed the continuous dynamic probabilistic condition monitoring algorithm.

No abstract