Digital Library

Machine learning in agile manufacturing; a usecase from offshore wind turbine product lifecycle management (plm) system

• Mehrnoosh Nickpasand
• Henrique M. Gaspar
• Hassan El Jaafari

(2024). ECMS 2024, 38th Proceedings
Edited by: Daniel Grzonka, Natalia Rylko, Grazyna Suchacka, Vladimir Mityushev, European Council for Modelling and Simulation.
DOI: http://doi.org/10.7148/2024
ISSN: 2522-2422 (ONLINE)
ISSN: 2522-2414 (PRINT)
ISSN: 2522-2430 (CD-ROM)
ISBN: 978-3-937436-84-5
ISBN: 978-3-937436-83-8 (CD) Communications of the ECMS Volume 38, Issue 1, June 2024, Cracow, Poland June 4th – June 7th, 2024

https://doi.org/10.7148/2024-0026

Mehrnoosh nickpasand, Henrique m. gaspar, Hassan el jaafari (2024). Machine Learning In Agile Manufacturing; A Usecase From Offshore Wind Turbine Product Lifecycle Management (Plm) System, ECMS 2024, Proceedings Edited by: Daniel Grzonka, Natalia Rylko, Grazyna Suchacka, Vladimir Mityushev, European Council for Modelling and Simulation. doi:10.7148/2024-0026

This paper describes how application of Machine Learning (ML) in Product Lifecycle Management (PLM) system results in reducing time of engineering change order (ECO), i.e., implementation of the design change’s effects, and thus, improving the responsiveness of the manufacturing system. A current challenge in PLM system is the long duration of engineering change process from when a component in product’s design is changed until when the change is reflected in the manufacturing operation in which the component is consumed. This paper proposes that application of supervised ML, using classification method and a basic neural network algorithm, on 3D graphic design of a product, helps PLM system to predict and locate the placement of the design change in corresponding manufacturing operation faster than the current manual status quo. The model-based structure of such application highlights the practical steps in data preparation, training process, and evaluation of the model in PLM system environment. Throughout the way, the feature, extracted from the digital format of the product design, i.e., 3D graphics, is labelled by its correspondence to the operation in which it is consumed, and sent through the neural network algorithm to train it over such correspondence. The satisfactory result of the evaluation is reflected by learning curve and confusion matrix. With use of data from offshore wind turbine’s (OWT’s) hub in Siemens Gamesa Renewable Energy’s (SGRE’s) PLM system, this study denotes the logic behind such application, the methodologies, and frameworks, which are reusable in any typical manufacturing system supported by PLM system. Although, for the utilization of the trained model, there are still some hurdles to overcome. The paper concludes with highlights of challenges, risks and limitations of the model helping other industries to predict potential challenges and realistically estimate the required budget for such implementation.