AI-PROFICIENT at the 16th World Congress on Engineering Asset Management in Seville, Spain, 5-7 October 2022
The 16th World Congress on Engineering Asset Management (WCEAM 2022) “Value – Centered And Intelligent Asset Management In The 4th Industrial Revolution Era” will be held from October 5th to October 7th, 2022, in Seville, Spain. WCEAM 2022 will involve experts in the application of techniques to achieve intelligent asset management in an industrial environment. The main premise of the congress’ theme is how to integrate intelligence into the asset management, through the management of value, knowledge and risk.
AI-PROFICIENT partner Fundación Tekniker will present the paper “Providing optimal initialization conditions for a complex extrusion process” during the congress’ session “Asset Management and Decision support systems” (14:30-16:00, 5th October 2022). In specific, Eider Garate from Fundación Tekniker will present the results on modelling and optimization related to the extrusion restart use case (CONTI-2) of AI-PROFICIENT project.
An extrusion is a continuous process in which the final products are obtained by shaping molten materials through a forming tool. This process is commonly used in the production of pneumatics, and it is considered a complex physicochemical problem which depends on the compound used, the desired final product, the operating conditions, and the internal state of the process. The adjustment and optimization of the extrusion process is an important issue to deal with because the errors and defects generated at this stage have a direct impact on the final product’s quality. The aim of the study, conducted by Eider Garate-Perez, Kerman López de Calle-Etxabe, Susana Ferreiro, Borja Calvo and Julien Hintenoch, is to construct a model which links extruder’s initial conditions with final product’s quality based on the analysis of a big volume of historical data. This research obtains a model based on controllable variables capable to predict the final product’s quality so that it could be used to understand the input space for a future optimal conditions’ search.