Sure! Here’s the translation into American English:
—
Hapag-Lloyd, one of the leading companies in global maritime transport, has taken an innovative step by developing a machine learning (ML) based assistant that enables more accurate predictions of arrival and departure times for its ships. With a fleet of over 308 modern vessels transporting 11.9 million TEUs annually, the company aims to optimize its scheduling, a crucial aspect of the industry that directly impacts the quality of service provided to customers.
Schedule reliability is measured as the percentage of ships that arrive within a calendar day of their estimated time of arrival, which the company communicates weeks in advance. Before implementing this technology, Hapag-Lloyd relied on statistical calculations and basic rules, which hindered its ability to adapt to changing real-time conditions, such as port congestion. This traditional approach required considerable manual intervention from the operations teams.
Developing this ML solution presented several challenges, particularly the need to account for the dynamic conditions of maritime transport. Factors such as weather, port delays, and unforeseen events requiring route changes complicated the process. A notable example was the blockage of the Suez Canal by the Ever Given in March 2021, which forced the diversion of numerous vessels and increased travel times by up to ten days.
Another challenge was integrating a large volume of historical data with real-time information sources. Hapag-Lloyd needs its model to operate at scale, encompassing 120 ship services and 1,200 unique routes. For this, MLOps infrastructure is crucial; it allows for continuous monitoring of the model’s performance and quick updates, ensuring that models are regularly retrained to adapt to changing patterns and maintain their effectiveness in real-time.
The old scheduling approach was insufficient to manage the complexities of modern maritime transport. Therefore, there was a need to create a comprehensive system that could handle predictions with infrastructure capable of supporting ML operations on a global scale.
Hapag-Lloyd operates an extensive network resulting in over 3,500 port arrivals monthly. Each vessel has a predefined itinerary, meaning arrival times must be estimated in advance to facilitate critical logistics. Any delays experienced at an intermediate port automatically affect arrival times at subsequent ports, leading to exhaustive rescheduling that can disrupt transshipment connections.
To forecast arrival times (ETA), Hapag-Lloyd utilizes internal data stored in a data lake, which includes detailed schedules for its vessels, port performance information, and real-time congestion data. The Automatic Identification System (AIS) is also integrated to accurately track ship movements.
The models developed from this information have shown promising results, demonstrating increased accuracy and a significant reduction in response times compared to traditional methods. This advancement competitively positions Hapag-Lloyd in the global market, marking a significant milestone in the maritime transport sector.
via: MiMub in Spanish