Multi-device state monitoring method based on federal learning

The invention provides a multi-device state monitoring method based on federated learning. The method comprises the following steps: collecting multi-node device data; carrying out data preprocessing, abnormal value cleaning and standardization; training a sub-node model; uploading the sub-models; fusing a central node model; issuing child nodes by the central node model; and performing equipment state prediction according to the training model. And when a new child node is added into the federated learning network, in order to reduce the time consumption of retraining, the center node issues the current model to the child node to be trained. According to the multi-device state monitoring method based on federated learning provided by the invention, a federated learning mechanism is introduced to improve the accuracy and generalization of multi-node model training, and meanwhile, the safety of training data is effectively ensured. 本发明提出了一种基于联邦学习的多设备状态监测方法，包括如下步骤：多节点设备数据采集；数据预处理清理异常值并标准化；子节点模型训练；子模型上传；中心节点模型融合