Data generation method for missing electroencephalogram signals

The invention discloses a data generation method for missing electroencephalogram signals, and belongs to the technical field of biological science. An electroencephalogram signal acquisition device is influenced by factors such as electromagnetic interference, disturbance of a subject and foreign matters on the brain scalp when recording electroencephalogram signals, and the acquired multi-channel electroencephalogram signals often have the condition of data missing. Therefore, in order to optimize low-cost acquisition equipment, data generation of missing electroencephalogram signals is realized. According to the method, a multi-modal information perceived graph is adopted to generate a learning model, a graph attention mechanism is combined, and spatial-temporal information enhanced electroencephalogram signal representation is learned for highly sparse discrete observation values; and multi-modal observation data related to corresponding acquisition equipment sensors and adjacent nodes thereof are reconst