Title: Fault monitoring and life prediction of bearings in automated production lines based on cloud-edge collaboration with optimised allocation of algorithm

Authors: Qian-Han Zhang; Qing-Hai Xie; Bing-Yan Wei; Tao Ma; Jin-Ping Du; Ying-Ming Shi; Jun-Xian Han; Lei Geng

Addresses: Hebei Institute of Mechanical and Electrical Technology, Xingtai City 054000, Hebei Province, China; Xingtai Technology Innovation Centre for Intelligent Production Line and Equipment, Xingtai City 054000, Hebei Province, China ' Hebei Institute of Mechanical and Electrical Technology, Xingtai City 054000, Hebei Province, China; Xingtai Technology Innovation Centre for Intelligent Production Line and Equipment, Xingtai City 054000, Hebei Province, China ' Hebei Institute of Mechanical and Electrical Technology, Xingtai City 054000, Hebei Province, China ' Hebei Institute of Mechanical and Electrical Technology, Xingtai City 054000, Hebei Province, China ' Hebei Institute of Mechanical and Electrical Technology, Xingtai City 054000, Hebei Province, China; Xingtai Technology Innovation Centre for Intelligent Production Line and Equipment, Xingtai City 054000, Hebei Province, China ' Tangshan Polytechnic University, Tangshan City 063299, Hebei Province, China ' Hebei Institute of Mechanical and Electrical Technology, Xingtai City 054000, Hebei Province, China ' Hebei Institute of Mechanical and Electrical Technology, Xingtai City 054000, Hebei Province, China

Abstract: This study develops a cloud-edge collaborative framework to optimise real-time monitoring and process control in automated production lines, focusing on the critical role of ball bearings. A comprehensive cloud-edge collaborative framework is developed to optimise task allocation between cloud and edge computing. At the edge level, an optimised deployment model for edge servers is established, taking into account computation time and workload balance to determine the optimal deployment nodes and required number of edge servers. For real-time bearing fault diagnosis, a lightweight hollow convolutional neural network (HCNN), which is designed to minimise latency while maintaining high diagnostic accuracy. On the cloud side, an enhanced HCNN model incorporating attention mechanisms and bidirectional long short-term memory (BiLSTM) units is employed for bearing lifespan prediction, improving predictive precision. The proposed framework is validated on an automobile pressure plate production line through real-time collection of rolling bearing signals. Experimental results confirm the effectiveness of the proposed approach in both fault diagnosis and lifespan prediction, demonstrating its feasibility for intelligent industrial applications.

Keywords: cloud-edge collaboration; bearing failure; prediction; hollow convolution.

DOI: 10.1504/IJAHUC.2026.155558

International Journal of Ad Hoc and Ubiquitous Computing, 2026 Vol.52 No.4, pp.229 - 242

Received: 17 Oct 2025
Accepted: 03 Dec 2025
Published online: 05 Aug 2026 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article