Shipping is missing out on real benefits of data analysis
Low-cost artificial intelligence and machine learning tools can unlock the value in data collected on board vessels
THE use of intelligent shipboard data has evolved at an astonishing pace over the past five years. We have moved from handwritten entries into the engine room logbook to artificial intelligence tools assessing and managing vessel machinery. Onboard operations have been transformed.
Innovations such as the cloud, cognitive analytics, the Industrial Internet of Things (IIoT), and advanced cybersecurity tools are improving vessel operations. As more tech-savvy executives step into leadership positions at ship owning and managing companies, and as engineers who grew up with high-tech gadgets occupy shipboard engine control rooms, the maritime industry is on a path to understanding the value of smart data collected onboard.
However, while there is broad agreement that technology trends are pointing to increased use of artificial intelligence, it’s clear that the full benefits of digitalisation are not understood. In a survey of maritime executives, only one in eight respondents reported using a predictive analytics tool — and none of them were using any form of artificial intelligence.
There are now inexpensive tools using artificial intelligence and machine learning available in the maritime sector that would enable management to access the full value of the data their vessels are already collecting. Even so, based on input SparkCognition has received from shipowners and ship managers, less than 5% of the data collected from sensors of shipboard machinery is analysed for performance analysis purposes. That analysis is gold to a data scientist who understands marine operations.
Insights gained from the terabytes of data being collected from shipboard machinery sensors present opportunities to prevent unscheduled downtime, optimise maintenance planning, and reduce capital replacement costs.
The challenge used to be how to collect data but today it is how to best distill that data into actionable guidance. Ships have a widely disparate set of machines with differing operating parameters. While much attention will be devoted to the more critical assets including marine propulsion units, generators, lubricating systems, and assets specific to a ship such as the cryogenic equipment on LNG tankers, a failure of a humble pump could have a considerable impact on the capability of the vessel.
Engineering management is eager to have the best toolbox available to provide deep understanding of all systems, and how anomalous behaviour (or failure) of one can have a major impact on the engineering plant somewhere downstream.
Artificial intelligence tools ingest operating data from shipboard sensors and classify it into normal behaviour or anomalies. Once this classification is done, an automated model-building system writes targeted rules that allow deeper insights into the machine’s operation. This tool provides a more comprehensive understanding of how the machine behaves in all its operating states.
In a recent case, the manufacturer of a critical asset believed that its machine had seven operating states, but automated model-building (combined with more operating data) found two additional operating states of which the manufacturer was unaware.
Natural language processing is another essential tool that is poised to make a difference in the maritime industry. This technology digests hundreds of equipment manuals, making them searchable and enabling human-to-text interactions. The information of this detail and quality allows managers to uncover new insights and make more informed engineering decisions. This results in decreased downtime (and off-hire) as sharing the data with in-voyage engineers, ashore managers, classification societies, and even ship repair yards enables a streamlined approach to reacting to suboptimal conditions before harm is caused to the machinery. Natural language processing tools also enable better forward planning for maintenance and capital replacement.
Given the role of AI in analysing data, detecting anomalies, predicting failure, and protecting assets with cognitive-based maintenance programs, it is rapidly becoming a requirement in the offices of forward-thinking vessel owners and managers. We’ve come a long way from relying on paper logbooks for data, and given the numerous areas where AI is already providing concrete returns, it’s exciting to see what comes next.
Walter Mitchell is maritime director of artificial intelligence company SparkCognition