Multi-value chain automobile part demand prediction method based on dynamic heterogeneous graph convolution

The embodiment of the invention provides a multi-value chain automobile part demand prediction method based on dynamic heterogeneous graph convolution, and the method comprises the steps: obtaining multi-value chain historical data related to a target manufacturer in a historical first time period, and constructing a dynamic heterogeneous graph according to the multi-value chain historical data; processing the dynamic heterogeneous graph into a plurality of static heterogeneous graphs according to unit time, and respectively obtaining an adjacent matrix and a feature matrix of each static heterogeneous graph; and inputting the adjacency matrixes and the feature matrixes of the plurality of static heterogeneous graphs into a trained dynamic heterogeneous graph convolutional neural network and long and short term memory network combined prediction model (DHGCNLSTM) according to a time sequence to obtain an accessory demand prediction result of the target manufacturer in a second time period in the future. Accor