Bibliometric Literature Review of Integrated Data and Model Based Diagnosis Approaches for the Industry 4.0

Abstract

The ubiquity of Cyber-Physical Systems and the Internet of Things has created a data-based environment that has allowed a significant number of applications linked to the "Industry 4.0" paradigm for the manufacturing industry. In this field, Artificial Intelligence is playing an important role for the development of automatic diagnostic tasks. This work carries out a review of the literature on the latest advances in Industry 4.0 around automatic diagnosis based on two specific objectives: (i) Determine the possible combinations between Data Based Diagnosis and Model Based Diagnosis, and (ii) Identify the hybridization opportunities and existing frameworks to integrate machine learning approaches to reuse solutions/algorithms for Industry 4.0 problems. The bibliometric review of the literature was guided by a simplified version of the PRISMA 2020 standard. Some of the main results of the review are that most of the combination schemes use observer-based approaches as the residuals generators. On the other hand, the largest number of machine learning applications for Industry 4.0 have been for smart manufacturing, focusing on failure management. As for the most used techniques, there are those of deep learning and ensemble methods.