Predictive Maintenance -- Bridging Artificial Intelligence and IoT

G.G. Samatas,S.S. Moumgiakmas,G.A. Papakostas

DOI: https://doi.org/10.48550/arXiv.2103.11148

2021-04-21

Abstract:This paper highlights the trends in the field of predictive maintenance with the use of machine learning. With the continuous development of the Fourth Industrial Revolution, through IoT, the technologies that use artificial intelligence are evolving. As a result, industries have been using these technologies to optimize their production. Through scientific research conducted for this paper, conclusions were drawn about the trends in Predictive Maintenance applications with the use of machine learning bridging Artificial Intelligence and IoT. These trends are related to the types of industries in which Predictive Maintenance was applied, the models of artificial intelligence were implemented, mainly of machine learning and the types of sensors that are applied through the IoT to the applications. Six sectors were presented and the production sector was dominant as it accounted for 54.54% of total publications. In terms of artificial intelligence models, the most prevalent among ten were the Artificial Neural Networks, Support Vector Machine and Random Forest with 27.84%, 17.72% and 13.92% respectively. Finally, twelve categories of sensors emerged, of which the most widely used were the sensors of temperature and vibration with percentages of 60.71% and 46.42% correspondingly.

Machine Learning,Artificial Intelligence,Networking and Internet Architecture,Systems and Control

What problem does this paper attempt to address?

The problem that this paper attempts to solve is the application trends and challenges of machine learning (ML) and Internet of Things (IoT) technologies in predictive maintenance (PdM). Specifically, through a systematic literature review (SLR), the paper explores the following issues:

1. **The relationship between PdM and ML**: It studies how PdM can be used in combination with ML to improve maintenance efficiency and reduce costs.
2. **The application fields of PdM**: It analyzes in which industries PdM has been applied between 2011 and 2020, especially in six major fields such as production, electronics, energy, transportation, commerce, and aviation/space.
3. **The methods and techniques of PdM**: It discusses in detail the ML methods used for PdM during this period, including artificial neural networks (ANN), support vector machines (SVM), random forest, etc.
4. **The types of sensors used in PdM**: It analyzes the most commonly used sensor types in PdM applications, such as temperature sensors, vibration sensors, noise/acoustic sensors, etc.

In addition, the paper also discusses some challenges faced by PdM in practical applications, such as data quality, unbalanced data processing, sensor accuracy, and the black - box characteristics of ML models. These challenges have an important impact on the practical application effect of PdM, so further research and improvement are required.

Through the above analysis, the paper aims to provide insights into the latest trends and future development directions of the combined application of PdM and ML for the industrial and academic communities.