Periodic multi-node power load prediction method and related device

The invention discloses a periodic multi-node power load prediction method and a related device, and the method comprises the steps: obtaining the historical load and influence factor data of a to-be-predicted load node and an adjacent node, and constructing a multi-node power load original data set; carrying out preprocessing to obtain a multi-node load sample data set, screening influence factors from the multi-node load sample data set, and constructing a final multi-node load sample data set; dividing according to workdays and holidays and festivals to construct a periodic multi-node load sample data set; fusing the graph convolutional network and a bidirectional gating cycle unit to build an initial periodic multi-node power load prediction model; and finally, optimizing hyper-parameters of the initial periodic multi-node power load prediction model based on the data set so as to determine a final periodic multi-node power load prediction model. The problem that in the prior art, only load prediction is