Failure prediction in production line based on federated learning: an empirical study

Data protection across organizations is limiting the application of centralized learning (CL) techniques. Federated (FL) enables multiple participants to build a model without sharing data. Nevertheless, there very few research works on FL in intelligent manufacturing. This paper presents results an empirical study failure prediction production line based FL. (1) designs Support Vector Machine and federated random forest algorithms for horizontal vertical scenarios, respectively; (2) proposes experiment process evaluating effectiveness between CL algorithms; (3) finds that performance are not significantly different global testing data, partial estimated unknown Bosch respectively. The fact data heterogeneous enhances our findings. Our reveals can replace prediction.