Work

Machine Learning in Smart Manufacturing: Challenges and Solutions

Year: 2026

Type: article

Abstract: In this paper, we consider the intertwined challenges that arise for process optimisation within smart manufacturing, reviewing fundamental aspects including concept drift in relation to the dynamic nature of manufacturing, the prevalence of heterogeneous data streams, integration of prior domain knowledge, and trustworthiness and transparency of machine learning models. We investigate the real-world nature of these challenges and the proposed solutions in the literature for each of the aforementioned challenges. Whilst other surveys exist for optimisation in the field of smart manufacturing, there is a lack of a comprehensive review addressing the underlying conceptual issues associated with the deployment and adoption of machine learning in smart manufacturing.

Source: Applied Sciences

DOI: 10.3390/app16178599

Authors: Omar Adalat, Nereida Polovina, Savas Konur

Institutions: Imperial College London, Manchester Metropolitan University, Aston University

Language: English

Topic: Digital Transformation in Industry

Subfield: Industrial and Manufacturing Engineering

Field: Engineering

Domain: Physical Sciences

SDG: Industry, innovation and infrastructure

Open Access status: gold

Estimated APC paid: $2,896