Inverse modeling for weight prediction from hydraulic data
Publication date
2026-05-07
Document type
Konferenzbeitrag
Author
Plambeck, Swantje
Schmidt, Maximilian
Wegner, Jannes
Wieck, Jan Christian
Fey, Goerschwin
Organisational unit
Hamburg University of Technology
Fraunhofer – Center for Maritime Logistics and Services (CML)
Publisher
Universitätsbibliothek der HSU/UniBw H
Book title
Machine learning for cyber physical systems : proceedings of the conference ML4CPS 2026
First page
101
Last page
110
Peer-reviewed
✅
Part of the university bibliography
Nein
Language
English
Keyword
Cyber-physical systems
Decision trees
Inverse modeling
Hydraulic system
Abstract
We present a case study on data-driven model generation for weight estimation in an Automatic Lashing Platform (ALP) using the Flowcean framework. The ALP is a cyber-physical system (CPS) that automates the insertion and removal of twistlocks during container handling at ports. The ALP operates using a self-sufficient, energy recovering hydraulic system, eliminating the need for an external energy supply. Accurate weight estimation of containers is crucial for optimizing valve adjustments in the ALP’s hydraulic system and maximizing energy recovery. We formulate the weight estimation task as an inverse modeling problem, where the goal is to predict container weight based on pressure curves measured in the hydraulic system. While models trained on raw pressure data achieve a Mean Absolute Error (MAE) of 0.815 tons, adding domain-knowledge and engineered features reduces the error to 0.293 tons. An ablation study shows that a lightweight, interpretable Decision Tree with a depth of 9 is sufficient to achieve an estimation error of approximately 1 ton. This validates that the approach is not only accurate but also suitable for deployment on embedded control hardware with limited computational resources.
Description
This contribution is part of the conference proceedings, which are licensed under the terms of the Creative Commons Attribution 4.0 International License (https://creativecommons.org/licenses/by/4.0/)
Version
Published version
Access right on openHSU
Open access