Equipment Fault Diagnosis Method Based on High-efficient Communication Federated Learning

Federated learning achieves joint training modeling of fault data from various factories while protecting privacy. However, due to the high heterogeneity of factory equipment operation data, traditional federated learning has low communication efficiency. To address the above issues, an equipment fault diagnosis method based on high-efficiency communication federated learning was proposed. Firstly, a federated dynamic weighted balancing model was proposed, which dynamically adjusts the training frequency and uploaded parameter amount of the factory sub end, and improves communication efficiency by reducing communication time. Secondly, a dual jump gate cyclic unit diagnostic model with an attention mechanism was proposed, which assigns different weights to different features to quickly extract fault features, effectively shortening communication rounds and improving communication efficiency. Finally, experimental validation was conducted using the bearing failure datasets of Case Western Reserve University and Paderborn University. The experimental results show that compared with the federated average algorithm, this method achieves a fault diagnosis accuracy of 92. 20% , while reducing communication time by 56. 88% and shortening communication rounds by 47. 37% , effectively improving communication efficiency.