Machine Learning and IIoT Application for Predictive Maintenance

Abstract

Advancement of technology is at large, and a lot of companies are exploring different ways to utilize it in the field of productivity and efficiency.

The fourth revolution in Industry, famously known as Industry 4.0 helped industries to widen its boundaries to support new sustainable business models for companies. After integrating Industrial Internet of Things (IIoT), Predictive Maintenance (PM) is seen as a potential sector that can take benefits for providing better sustainable solutions with increased insight in the manufacturing process. Machine Learning (ML) process/tools provides us a lot of meaningful information by analysing the raw data gathered by IIoT tools such as smart sensors and actuator. Utilizing the information/readout coming from different IIoT tools, performance and/or health of an equipment can be predicted with the help of Artificial Intelligence. This paper discusses the use of ML with IIoT to achieve PM from a data analysis perspective. Different traditional ML and deep learning techniques are discussed and compared with its potential use in predictive maintenance. A case study on data from a heat exchanger is used to demonstrate the advantage of deep learning in data analysis via comparison between neural networks (LSTM), ARIMA, and traditional machine learning (SVM). The results show a good performance of the deep learning LSTM model prediction compared to other techniques used in time series forecasting.