INFORM has published a new paper as part of the latest e-Journal of Ports and Terminals covering how machine learning (ML) is changing the ports and terminals around the world.
Published in Edition 83, the paper references a previous technical paper INFORM wrote last year for PTI on artificial intelligence (AI) and it was making its way into the terminal industry.
It stressed that the central question wasn’t how to prepare for AI but rather how the potential of ML can be built into core AI solutions.
Machine Learning is one of the topics that will be discussed at PTI's CTAC event in London between 6-8 May, 2019
The latest paper, titled ‘Machine Learning in Terminal Operations’ by Dr Eva Savelsberg, Dr Ulrich Dorndorf and Matthew Wittemeier, explores this idea.
View INFORM's solutions for the port and terminal industry by visiting their page in PTI's AIS Portal
In doing so, it details the methodology and results of an assessment INFORM undertook in order to explore ML’s potential to boost port operations.
The assessment, as the new paper says, looked to answer two key questions, the first asking if broader ML algorithms currently being used in other sectors be a crossover into maritime.
The second asked if they could be applied to real-world terminal data and can improvements be made to parameters that affect the optimization calculators of INFORM’s add-on Optimization Modules.
A particular point of the examination was container dwell time, a central variable in optimizing the placement of containers in a terminal.
In order to gauge potential improvements, the assessment acquired expected departure data from containers on ships, the railway, trucks and feeders.
It used INFORM’s optimization module to draw up an empirical model to determine the mean dwell time for containers with no expected departure time upon arrival.
Future assessments, INFORM goes on to say, should aim to review data from 2018 and run it against the same process to assess whether the findings from current findings are accurate or if there potential alternative patterns.