MSE 2024
Lecture
26.09.2024 (CEST)
Empowering Manufacturing through Adaptive Experimental Design and Machine Learning
BN

Dr. Bogdan Nenchev

Intellegens Limited

Nenchev, B. (Speaker)¹; Strickland, J.¹
¹Intellegens Limited, Cambridge (United Kingdom)
Vorschau
16 Min. Untertitel (CC)

In the dynamic landscape of modern manufacturing, the need for accelerated development takes centre stage. A transformative approach, seamlessly blending collaboration and innovation, emerges as a catalyst for swift product ideation, testing, and market release. Intellegens, leveraging its proficiency in managing sparse and noisy experimental data, introduces a collaborative model designed to reshape traditional paradigms.

In the manufacturing industry, especially in complex processes like casting single crystal Ni-based superalloys for aerospace applications, a significant focus is placed on the collection and analysis of structured data. This typically encompasses detailed information such as alloy chemistry, process parameters, and temperature evolution. These structured data points are crucial for understanding and controlling the manufacturing process, as they provide a clear and organised framework for analysis.

However, alongside structured data, there also exists a substantial amount of unstructured data, including timeseries and microstructure images. These unstructured data types present unique challenges in terms of processing and interpretation, due to their less organised nature. The combination of both structured and unstructured data types is essential in addressing the inherent challenges of these manufacturing processes. This accentuates the need for an innovative approach that can efficiently navigate this complex manufacturing space, aiming to reduce time and material consumption.

This research introduces a transformative manufacturing approach for bridging the gap between structured and unstructured data, machine learning, and adaptive Design of Experiments (DoE). The system's innovation lies in its capacity to recommend optimised production parameters for a new part, leveraging insights from cumulative learning derived from all preceding casts and components. By harnessing the capabilities of machine learning and adaptive DoE, the system efficiently streamlines and automates large-volume component production, ensuring increased output quality and reliability.

Intellegens highlights the transformative capacity inherent in this collaborative paradigm, challenging established conventions in product development. This study invites industry peers to actively engage in this endeavour, collectively reshaping the framework for efficient and innovative product creation. Our joint efforts aim to forge a future wherein manufacturing transcends its traditional boundaries, evolving into a dynamic collaboration situated at the confluence of expertise and technology, propelling the field towards unparalleled advancements.


Abstract

Abstract

Erwerben Sie einen Zugang, um dieses Dokument anzusehen.

Ähnliche Inhalte

© 2026