Hybrid Deep Learning-Based Predictive Maintenance Framework for Industrial Internet of Things Systems

Abstract

The widespread introduction of the Industrial Internet of Things (IIoT) has allowed smart sensors to be connected to industrial equipment, which means their working state can be monitored at all times, resulting in an increasing amount of operational data that could be utilized for predictive maintenance using intelligent algorithms. The traditional maintenance methods such as corrective maintenance, preventative maintenance can cause unnecessary maintenance expenses, unforeseen machine breakdowns, and lower productivity because of the lack of accurate prediction of machine degradation. In order to address these problems, this paper suggests a Hybrid CNN–LSTM framework for predictive maintenance combining Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks. The proposed framework consists of the following steps: industrial sensor data acquisition, data preprocessing, development of a hybrid deep learning model, and intelligent fault prediction. The model was tested with the AI4I 2020 Predictive Maintenance Dataset and was compared to the traditional machine learning and deep learning models, such as Support Vector Machine (SVM), Random Forest (RF), CNN, and LSTM. The performance was evaluated based on the following metrics: Accuracy, Precision, Recall, and F1-score. The experimental results showed that the proposed Hybrid CNN–LSTM framework achieved an Accuracy of 97.12%, Precision of 96.74%, Recall 96.39% and F1-score of 96.56% compared to the baseline models with stable convergence during training. The obtained results show that the suggested framework is accurate and reliable solution to intelligent predictive maintenance and early fault diagnosis in Industrial Internet of Things environment.