A Federated Learning Framework for Privacy-Preserving Predictive Maintenance in Distributed Smart Manufacturing

Predictive maintenance has become an essential component of smart manufacturing systems because it enables early detection of machine failures and reduces unexpected pro-duction downtime. However, conventional predictive maintenance approaches typically rely on centralized data collection and model training which may raise concerns regarding data privacy, communication overhead and data ownership in manufacturing environments. To address these challenges, this research proposes a privacy-preserving collaborative federated learning framework for predictive maintenance that can be deployed in distributed smart manufacturing systems. The proposed approach allows multiple factories to jointly train a machine failure prediction model without sharing raw data. In the framework, each factory trains a local multilayer perceptron (MLP) model using its own machine operational data, while a central server aggregates local model parameters using the Federated Averaging (FedAvg) algorithm to construct a global predictive model. The proposed framework was evaluated using the publicly available AI4I 2020 predictive maintenance dataset where multiple factories are simulated by partitioning the dataset into distributed clients. Experimental results show that the federated learning model achieves performance comparable to centralized machine learning baselines, reaching an accuracy of 99.93%, precision of 1.000, recall of 0.980 and F1-score of 0.990 while still preserving data privacy and IP protection across distributed participants.