Artificial Intelligence applied on the basis of powering of vessel and its

operational and management aspects

ChEng.Amit Singh,Sarthak Saxena
(AToS Remote Soulutions & IIT-Guwahati4 rth Year Mech Dept)
(343Clementtown Dehradun 248002 )
{amit4766@yahoo.co.in ,sa19062000@gmail.com}

System that can dictate the powering of vessels, onboard

Abstract - Oceans are consequential provenance of
activities and security. We enlisted that an AI-controlled
resources, transportation and bio-diversity on a grand
vessel operation, communication with low latency with
scale. Humankind uses oceans for various motives such
as haulage, victuals, trade, mineral extraction, power Coast and satellite, high-resolution sensor data
generation, warfare, sustention and leisure. Many of apprehension in real-time and marine traffic management
these activities fabricate marine pollution and straining will surely empower maritime routes. Eventually, we state,
in sustaining aquatic life, despite several protection how to implement these concepts in genuine.
barriers.
We are to harness some of the clamant aspects of AI-
controlled systems on power management and innovative
paradigms including certain recent applications and
references on AI.
Keywords – AI, Power Management, Power Generation

INTRODUCTION-

Vicenç Torra noted that the first definition of AI was in a

document prepared by J. McCarthy, M. Minsky, Figure1.Definition and Domain under AI
N. Rochester and CE Shannon for the meeting held in
Dartmouth (USA) during the summer of 1956 in which
the term ‘Artificial Intelligence’ was introduced. The In shipping industry, AI can incorporate
document elucidates AI as building a machine that carries Remote inspection, Fuel efficiency, Navigation and
out human-like behavior and would be called intelligence. awareness, Predictive and efficient power management
Human behavior is, however, not the only definition. system, etc. Automation, machine learning and AI
Other definitions are: cannot be limited to make operations smoother, efficient
and economical, but they can also help generate a lower
1. Act like people- McCarthy's definition. The Turing environmental footprint and create a higher quality-life
Test (1950) also uses this corresponds to human
for seafarers. Big data analytics will allow the industry
behavior. Example- Eliza system, a natural language bot
to uncover insights, trends, and correlations currently
(software program).
hidden in the shipping industry energy management,
2. Reason like people- The prominence here is the route planning and optimization, predictive
reasoning and not the result. The proposal is to develop maintenance, environmental management, as well as
systems that reason similarly to people. Cognitive vessel security.
science uses this standpoint. However, additional sensors and equipment to
3. Reason rationally- Here, Logic allows the measure the efficiency of operations promotes the big
formalization of reasoning. question of how to turn gargantuan datasets into real
vessel performance-enhancing solutions.
4. Act rationally- Again, the pursuit is the results, but now
evaluated objectively. e.g., in a game like chess, the Applications of AI in different parts of vessel
program aims at victory. Calculating the outcome is
irrelevant for achieving this goal. “Eco Marine Power, a Japan based provider of vessel
AI can enhance shipping routes, intonate vessel renewable energy system has announced that it is to begin
performance and power systems while reducing costs and incorporating AI into a range of its ship related technology
risk to mankind under the aegis of projects, specifically using the Neural Network Console
provided by Sony Network Communications

Deep learning, in the neural network console refers to a

form of machine learning that uses neural networks
modelled after the human brain, which Eco Marine Power can strengthen onboard-communication system, reflect
(EMP) says will offer it high versatility in application operating data to main deck, and maintain control
development across a wide variety of fields, including systems in ports and several areas during the journey.
signal processing and robotics. Initially company focuses 3. Crew and facilities management system –
on studying, how the Neural Network Console and AI can
assist with the development of an automated control It is necessary to maintain a healthy environment of a
system for its Energy-Sail system. vessel. A robust AI management system can determine
the best-course of management of the crew security,
Many companies alike have started developing gym, TV-room, mess, other facilities and crew’s social-
AI and a sophisticated data analytics section to conditions utilizing sensors, CCTV cameras, regular
incorporate in vessels. AI technologies are being used to monitoring system, etc. for a profitable shipping.
mimic human perception and cognitive abilities such as
seeing, hearing, reading and interpreting sensor data A big data-pool is needed to analyse all that
which are benefiting user interfaces aboard ship as and AI can easily determine a deep-neural-network
speech recognition directly helps in controlling types of based model to provide critical information and
equipment onboard that can drastically improve the management data to the crew and develop an
management and maintenance of power systems abroad economical and easy-managing ship control network.
vessel. Some areas of development in powering of Remote access via AI can give the current condition and
vessels are: data to a mainland-management-system reducing overall
powering and fuel usage.
1. The power management of vessel based on AI
4. Faster and economical delivery routes for retail shipping.
The sensors onboard the ship and the information from 5. Predictions on estimated times of arrival for ships
satellite can be used in data analysis software which in- as well as spotting trends and risks in shipping lanes
turn uses AI to stabilize the ship and determine the and ports.
route. 6. Analysis of historical shipping data by
considering factors like weather patterns and
AI actually resolves the data and maintain the speed and shipping seasons.
power distribution in the propellers (which can move 7. Using analytics to better understand customer’s
freely in a 3D plane) to move the vessel in appropriate challenges enable business processes to transform
direction. It also can check the wind, wave and that data to anticipate future.
temperature data to produce and distribute power to 8. Prevent halting shipping operations in natural
various systems economically. The systems on which AI disasters, terror attacks, etc.
can be applied – 9. Planning and optimizing of container terminal
operations, be it minimizing and automating exceptional
I) The ship-navigation system (includes propeller, case handling, predictive maintenance, and supply chain
radar, on-board sensors, maintaining the center optimization in terminals, ships, road transportation and
of gravity). warehousing
II) Ship power generation system
III) Ship auto-maintenance system Drones and robotics-machine to control the auxiliary
systems of different-types of ships. Sea-mining and survey-
AI can also determine appropriate route, speed to vessels management and many more example of AI based
maintain ocean traffic and load-to-power-usage ratio for applications –
running the ship most economically.
 SailRouter uses AI-based cloud application that
2. Ship operation control based on AI – helps ship-owners to reduce fuel consumption
and maintenance costs.
During the voyage, AI can maintain the distribution of
 VesselBot uses AI-based digital chartering
power onboard to any rooms, decks or other sections.AI
marketplace for the bulk maritime industry
can determine each sailor’s usages capacities and
determine the best course to save power by developing a
Proposed Method
neural-model.

AI can also develop a local power distribution model

depending on regular activities and sensor data. During 1. An integrated control system SEEC -
emergencies, AI can determine the suited actions. It also
The naval industry has been conventional and appears
 system-model uncertainties, sensor noise and fault
conditions and complex parameter interactions. These
MI techniques are often based on statistical analysis and
intelligent data handling algorithms.

1.B.- MACHINE INTELLIGENCE (MI)

1.B.1. - Data Flow Path

The data flow chart of ship performance and navigation

information is displayed in Figure 2. An appropriate MI
application is introduced and elaborated in each step of
the flowchart. Aforetime, performance and navigation
data is collected from various onboard sensors and data
Figure2. Data flow chart in ship performance and navigation information acquisition systems. Then, the data is further transferred
through a data pre-processing step. This pre-processed
data is communicated through data transmitters (onboard
the vessel) in smaller improved data sets. The same data
To be at the tail of implementation of improvements that
sets are obtained by shore-based data centers through
have already matured in other industries. The naval
data receivers. Then, these data sets are handled through
industry is no friend of risks, especially when the simple
a post-processing step and accommodated in data
fact of setting boats is a huge risk in itself. In this aspect,
storage facilities.
an AI-powered control-based “ship energy efficiency and
control (SEEC)” can revolutionize current maritime-ship The pre-post processed data used in the shipping
operations and control systems. The manufacturing, industry supports decision-making factors including
operation control and working principle of the system energy efficiency and system reliability. The energy
described- efficiency applications consist of identifying vessel
operating conditions to reduce overall fuel consumption.
The system reliability applications identify the health
1. A. - PERFORMANCE & NAVIGATION DATA conditions of onboard systems and determine optimal
maintenance measures to reduce the vessel's operating
1.A.1. - Integrated Bridge Systems
costs. The MI applications are demonstrated under a
The automation systems consist of a power management ship performance and navigation data set that are
architecture for engine and propulsion control systems obtained from a selected vessel (i.e., a bulk carrier) with
with respect to various engine room operations. the following particulars:
Additional units of bilge and ballast control, HVAC and
Ship length: 225 (m), beam: 32.29 (m), gross tonnage:
alarm & monitoring systems can also be a part of these
38.889 (tons), deadweight at max draft: 72.562 (tons).
automation systems. Thus, this system assists in
The vessel is powered by 2 stroke ME with maximum
obtaining a comprehensive overview of ship
continuous rating (MCR) of 7564 (kW) at the shaft
performance and navigation information to facilitate the
rotational speed of 105 (rpm). Furthermore, the vessel
execution of efficient navigation strategies.
has a fixed pitch propeller diameter 6.20 (m) with 4
1.A.2. - Data Handling blades (Perera et al., 2015b).
The speed-power plot with respect to relative wind
A cost-effective data handling solution is developed that conditions is presented in Figure 3.
consists of referential steps to avoid large-scale data sets
in real-time. Hence, a serialized data analytics is
developed using the subset of the large-scale data set to
overcome internal and external data handling issues.
These analytics can be used in real-time under IBSs to
improve respective data handling.
1.A.3. - Data Analytics

Machine intelligence (MI) based data analytics are

proposed to overcome data handling challenges due to
 the visibility of ship performance and navigation data.
1.B.1. Data Compression and Expansion

Data compression and data expansions are the last and

first steps in data pre-processing and data post-
processing respectively. Another machine learning auto
encoder network approach governs these steps. Auto
encoder is an unsupervised learning method with a feed-
forward neural network, also categorized as deep
learning approach. The main objective of an auto
encoder is to recreate the input at the output of the neural
network using compression/expansion of data sets.

1.B.2. - Data Integrity Verification

The next step of post-processing is integrity verification.

A subset of performance and navigation data is always
transferred by ships as automatic identification system
(AIS) messages. Similarly, that information is
exchanged electronically through other ships; AIS base
stations, and satellites nearby. This information can be
obtained by shore-based data centers to improve the
integrity of ship performance and navigation data.

 The last step is data regression. The estimated

data points help to calculate the respective
parameter values. This step may consist of
various algorithms to reduce the fluctuations in
the estimated data. Similar models can be
utilized to evaluate energy efficiency under
Figure3.Mod Speed Power Plot with PCA various operational conditions.
The respective speed power data set is standardized,
where the mean values are subtracted from each  Primarily, SEEC develops an appropriate
parameter and the variance values are set to 1.0 (i.e., methodology to handle large-scale data sets.
each parameter with equal variance). It's observed from This model proposes a marine engine-
the speed-power plot that the speeds decreases due to centered data flow chart to handle such
high relative winds for the same engine power levels. large-scale data sets as a big data solution
along with improving the quality of
Low-speed situations are removed from this data to
respective navigation strategies.
improve the visibility of ship performance and
navigation information.

1.B.2. Sensor Fault Detection

The pre-processing step is further divided into three

sections: sensor fault detection, data classification and
data compression. This step involves detecting sensor
faults and removing erroneous data regions from the
performance and navigation data set by observing the
mean and variance of each parameter.

1.B.3. - Data Classification

Data classification is the next step of the data flowchart.

An engine-centered data classification approach
develops navigation strategies. Accordingly, the large-
scale data sets are classified into subsets concerning Figure4. NAI (Similar to SEEC) Behaviour Tree
engine operating regions. The small data sets enhance
 2. Power management system based on SEEC - overall operations of a vessel. AI can initialize an end-
to-end demonstration of a digital certification and
SEEC's data set can maintain the power generation and endorsement process utilizing a digital repository for
output of the engine and consequently, auto- manage the verified crew documentation, training logs and an
vessel's speed and direction. SEEC can determine the approval system. AI can introduce the potential to
convenient route, speed and load-to- power-usage ratio significantly improve the transparency, authenticity and
for economic voyages. ease of working with crew certification; both for the ship
A cognitive model based on SEEC i.e. the “Engine owner/operator and, most importantly, for the crew
module” can be prepared for power and control themselves.
mechanisms.
For example - Recent utilization and experiments of AI in maritime
industry
A) AI to predict the best fuel-efficient way
to operate a vessel. Innovation in the maritime industry has been a gradual
B) In windy or high tide conditions, the propulsion process. Many companies and research institutes are
system can be organized via calculating conducting studies and experiments on AI-based
physical parameters and running them through management of maritime vessels. Examples of such
a cognitive AI model. experiments are –

1. Maana, a US Based Start-Up which has

developed a knowledge platform which enables
3. Ship operation control module –
companies such as Shell and Chevron to
This section contains various controlling modules based encode their human expertise into a
on SEEC. For each module, a singular sub-data set is computational graph.
prepared that auto-determines the best course of 2. Kalmar also plans to deploy AI through a
parameters under various conditions. Such module can model that calculates fuel consumption for Eco-
include - Reachstacker based only on three inputs from a
customer regarding cargo handlingpatterns.
a. Auto-Navigation module (auto-pilot ship 3. Rolls-Royce launched a state-run car ferry
depending upon sea, weather and cargo based on AI that avoided obstacles on a 1-mile
conditions). route. These systems help the ship perform
b. Auto-Maintenance module (ship maintenance easily in rough winter, handling snow and
via scanning ship and notifying sailors). winds..
c. Auto-Docking module (for docking at port 4. Wärtsilä subsidiary Transas’ Navi-Planner uses
and cargo loading). machine learning to optimize voyage planning.
d. Maritime surveillance module (for Safe navigation routes are automatically
robust surveillance ships). derived based on the latest charts and
e. Predictive analytics module (predicting environmental information. It records any near-
vessel propulsion failure, toxic blooms in misses and other incidents during voyages and
coastal waters, etc. for environmental and adapt routes and speeds to ensure scheduled
economic sustainability). arrivals.
f. Berthing and cargo module. 5. The Singapore Port and IBM collaborated on a
project called SAFER, which uses a machine
learning approach to predict arrival time and
unusual behavior of vessel, traffic hot spots,
and illegal bunkering.
6. Swedish ferry operator Stena Line reduces fuel
consumption by 2-3 percent per trip by using
AI to predict the most fuel-efficient ways to
operate a vessel. The project began in 2018
onboard Stena’s Scandinavica vessel on the
Gothenburg – Kiel route. The AI software,
Figure5. Naval Artificial Intelligence in the form of Tree Stena Fuel Pilot, will be installed on five more
Stena Line vessels.
A cognitive AI computer system is not enough for
Other research modules can also be found that deals with
Digitalization and Innovation in maritime transport,
Applications of big data from AIS, Maritime surveillance,
Environmental and economic sustainability, Energy
efficiency, Predictive analytics, etc.

Various underwater ROVs are also being developed. With

extended and convenient operation, ROV gets energy,
transmission data and control commands by umbilical
cables and is widely applied to resource exploration,
laying of submarine cables, marine scientific research,
etc. The major maritime powers worldwide have created
various ROVs that handle variegated tasks and dive
depths, such as the Japan KaiKo ROV that set a world
record by diving 10,000 meters to the Mariana Trench.

The future transport technology will be dominated by

autonomous vehicles. Though maritime transport is the
most conservative sector, it will not avoid general trends
and in near future, autonomous vessels will command the
oceans.

CONCLUSIONS

The study provides a pervasive review of big data and

AI in the maritime domain, combining bibliometric
analysis with systematic content analysis and the
proposal of new models. The fact that an AI-controlled
vessel operation, communication with low latency with
coast and satellite, high-resolution sensor data
apprehension and marine traffic management will
empower the seaways is embraced. Eventually, we
deliver the verity on how to implement these concepts in
the real world.

ACKNOWLEDGMENTS

The authors are thankful for studies and research papers

published in the domain of maritime innovation and
development and express their courtesy to the extant
maritime authorities for some included data sets.

REFERENCES
[1] E. Kulbiej and P. Wołejsza “Naval Artificial Intelligence” Aug
2020.
[2] Heller, Christian H. (2019) "The Future Navy—Near-Term
Applications of Artificial Intelligence," Naval War College
Review:
Vol. 72 : No. 4 , Article 7
[3] Perera, L.P. and Mo, B. (2016a),” Machine Intelligence for Energy
Efficient Ships: A Big Data Solution” 143-150. (Proceedings of the
3rd International Conference on Maritime Technology and
Engineering (MARTECH 2016), Lisbon, Portugal, July, 2016)
[4] Yang Yo and Zhenxing “The Application of Artificial Intelligence in
Ocean Development : In the View of World Expo 2010”