A Novel Multi-Objective Fuzzy Deep Learning Framework for Predictive Maintenance in Industrial Internet of Things
As industrial systems advance within the framework of Industry 4.0, predictive maintenance (PdM) has become essential to minimize downtime and improve operational efficiency in industrial Internet of Things (IIoT) environments. However, traditional PdM methods often struggle with handling complex hi...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10909436/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | As industrial systems advance within the framework of Industry 4.0, predictive maintenance (PdM) has become essential to minimize downtime and improve operational efficiency in industrial Internet of Things (IIoT) environments. However, traditional PdM methods often struggle with handling complex high-dimensional data and managing uncertainty, leading to limitations in predictive accuracy and adaptability. This paper introduces MO-FCNN, a novel PdM model that integrates fuzzy logic with deep convolutional neural networks (DCNN) to manage data uncertainty and utilizes non-dominated sorting biogeography-based optimization (NSBBO) to improve the overall maintenance decision-making process. Building upon the MO-FCNN-based PdM framework, this approach targets the unique challenges of high-dimensionality and dynamic data in IIoT, enhancing both the interpretability and performance of PdM systems. The proposed method leverages NSBBO in two critical areas: backpropagation optimization within CNN’s fully connected neural network layers and membership function optimization to refine fuzzy logic handling of uncertain or noisy information. This combined framework is designed to maximize equipment uptime and optimize resource allocation by dynamically adapting to varying maintenance demands and operating conditions. Extensive experimental evaluations, including performance metrics on resource allocation, predictive accuracy, and maintenance scheduling efficiency, demonstrate that MO-FCNN outperforms traditional PdM approaches. The findings underscore the ability of this model to serve as a reliable tool for advanced asset management, offering significant improvements in predictive accuracy, computational efficiency, and adaptability to various IIoT conditions. |
|---|---|
| ISSN: | 2169-3536 |