Variations in Maritime Vessel Alert Systems
Pollari, Eikka (2018)
Pollari, Eikka
2018
Automaatiotekniikka
Teknisten tieteiden tiedekunta - Faculty of Engineering Sciences
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2018-12-05
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tty-201811122564
https://urn.fi/URN:NBN:fi:tty-201811122564
Tiivistelmä
The maritime industry is continuously raising the level of automation on board vessels, the eventual goal being to introduce remotely operated and autonomous commercial vessels in the near future. Due to this development, more intelligent vessel alert systems are needed, as they are required to provide more detailed information and sophisticated alarm management functions. However, developing an intelligent alert system customisable for various vessel types and/or implemented as a retrofit installation is a challenging task. This thesis aims to help in that work.
The objective of this thesis is summarised into three goals. The first one was to find out what kind of challenges the variations in alerts and alert systems between various vessels and vessel types cause for developing an intelligent alert system customisable for variable types of vessels. Based on the literature review and the carried out interview study, the variations complicate the required grouping of alerts when designing various alarm management functions: due to differently formed alerts, grouping is automatically done very difficult and manually very time-consuming and demanding.
The second goal was to suggest a development roadmap for the intelligent alert system so that the vessel types with less variations come first. According to the interviews, the amount of variations can be derived from the complexity, i.e. the operational purpose and automation level, of the vessel type and vessel. Thus, the simplest vessel types, such as tugs, oil tankers and containers were suggested to be focused on first. The interviews suggested also that the simplest vessel types should be forgotten and the development should be started from a bit more complex vessels, e.g. from off-shore service vessels.
The third goal was to provide examples of how to take the variations into account when developing the intelligent alert system. For this, an Excel tool to categorise automatically differently formed alert signals of signal lists was developed by studying and utilising real-life alert lists and by applying methods of text classification. Also, grouping methods of the SFI Group System were utilised.
In tests the developed signal list categoriser proved to be effective and easily updatable for better accuracy and more comprehensive categorising capability. Thus, the goals set for this thesis were reached. If more detailed data about the variations in alerts and alert systems were wanted, a very large amount of alert list data should be gathered and statistically analysed. However, that would be a very challenging study to accomplish due to the amount of needed data and the difficulty to acquire it. The developed signal list categoriser could be further developed e.g. by analysing more alert lists and applying various, sophisticated machine learning algorithms utilised in text analytics.
The objective of this thesis is summarised into three goals. The first one was to find out what kind of challenges the variations in alerts and alert systems between various vessels and vessel types cause for developing an intelligent alert system customisable for variable types of vessels. Based on the literature review and the carried out interview study, the variations complicate the required grouping of alerts when designing various alarm management functions: due to differently formed alerts, grouping is automatically done very difficult and manually very time-consuming and demanding.
The second goal was to suggest a development roadmap for the intelligent alert system so that the vessel types with less variations come first. According to the interviews, the amount of variations can be derived from the complexity, i.e. the operational purpose and automation level, of the vessel type and vessel. Thus, the simplest vessel types, such as tugs, oil tankers and containers were suggested to be focused on first. The interviews suggested also that the simplest vessel types should be forgotten and the development should be started from a bit more complex vessels, e.g. from off-shore service vessels.
The third goal was to provide examples of how to take the variations into account when developing the intelligent alert system. For this, an Excel tool to categorise automatically differently formed alert signals of signal lists was developed by studying and utilising real-life alert lists and by applying methods of text classification. Also, grouping methods of the SFI Group System were utilised.
In tests the developed signal list categoriser proved to be effective and easily updatable for better accuracy and more comprehensive categorising capability. Thus, the goals set for this thesis were reached. If more detailed data about the variations in alerts and alert systems were wanted, a very large amount of alert list data should be gathered and statistically analysed. However, that would be a very challenging study to accomplish due to the amount of needed data and the difficulty to acquire it. The developed signal list categoriser could be further developed e.g. by analysing more alert lists and applying various, sophisticated machine learning algorithms utilised in text analytics.