Residential electricity consumption prediction method based on deep learning and federated learning

The invention belongs to the technical field of electric quantity prediction, and particularly relates to a resident electricity consumption prediction method based on deep learning and federated learning, and the method comprises the following specific steps: 1, collecting user data; 2, constructing an edge side training model based on federated learning; 3, decomposing a data set; 4, establishing a federal learning depth model; 5, parameter training is carried out; and 6, obtaining available model parameters of each edge side model through edge side training, updating and optimizing all the edge side models by receiving encryption model parameters of the edge side, and realizing aggregation of the edge side models at the cloud. An FL side model used for REC prediction is constructed based on EMD-LSTM, personalized processing is performed on the FL, and errors of REC prediction are reduced. 本发明属于电量预测技术领域，具体为一种基于深度学习和联邦学习的居民用电量预测方法，该基于深度学习和联邦学习的居民用电量预测方法的具体步骤流程如下：第一步：采集用户数据；第二步：基于联邦学习构建边缘侧训练模型；第三步：分解数据集；第四步：建