【摘要】 文中提出了一种基于深度学习以及联邦学习的电动机故障诊断模型。使用Transformer模型对电动机的运行数据进行分析以及故障诊断分类，在此基础上使用FedProx联邦学习算法，在多个客户端上使用不同的数据训练模型，并将训练后的模型上传到中央服务器进行聚合，使用聚合后的模型对设备进行故障诊断。实验结果表明，所提出的模型具有良好的性能，对数据的故障分类准确率满足电动机故障诊断的要求，同时联邦学习的方法有助于模型获得更多的数据特征，使得模型可以更好地进行故障诊断，同时对保护数据隐私也有一定作用。更多还原

【Abstract】 A motor failure diagnosis model based on deep learning and federated learning is proposed. Transformer model was used to analyze the running data of motor and classify the failure diagnosis. Fed Prox federated learning algorithm was adopted, different data were used to train the model on multiple clients before the trained model was uploaded to the central server for aggregation, and the aggregated model was used to carry out failure diagnose. The results show that the proposed model has good performance, and the failure classification accuracy of data can meet the relevant requirements of motor failure diagnosis. In addition, federated learning helps the model to obtain more data features, improve the quality of model failure diagnosis, and also plays a certain role in protecting data privacy.更多还原